Methods of Analysis by The U.S. Geological Survey National ...Inorganic Constituents and Method, Lab, and Parameter Codes Constituent Method Lab code Parameter and method code Mercury,

Methods of Analysis by The U.S. Geological SurveyNational Water Quality Laboratory—Determination ofOrganic Plus Inorganic Mercury in Filtered and UnfilteredNatural Water with Cold Vapor–Atomic FluorescenceSpectrometry

Water-Resources Investigations Report 01-4132

U.S. Department of the InteriorU.S. Geological Survey

Methods of Analysis by the U.S. Geological SurveyNational Water Quality Laboratory—Determination ofOrganic Plus Inorganic Mercury in Filtered and UnfilteredNatural Water with Cold Vapor–Atomic FluorescenceSpectrometry

By John R. Garbarino and Donna L. Damrau

_______________________________________________________

U.S. GEOLOGICAL SURVEY

Water-Resources Investigations Report 01-4132

Denver, Colorado2001

U.S. Department of the InteriorGale A. Norton, Secretary

U.S. Geological SurveyCharles G. Groat, Director

The use of trade, product, or firm names in this report is for descriptivepurposes only and does not imply endorsement by the U.S. Government.

For additional information write to: Copies of this report can be purchased from:

Chief, National Water Quality Laboratory U.S. Geological SurveyU.S. Geological Survey Branch of Information ServicesBox 25046, Mail Stop 407 Box 25286Federal Center Federal CenterDenver, CO 80225-0046 Denver, CO 80225-0286

Contents III

CONTENTS

Abstract 1Introduction 1Analytical method 2

1. Application 22. Summary of method 23. Sample collection and storage 34. Quality-control samples, sampling methods, sample-collection equipment,

and cleaning procedures 35. Contamination and interferences 46. Apparatus and instrumentation 57. Reagents and calibration standards 58. Sample preparation 69. Analytical procedure 7

10. Calculations 711. Reporting of results 7

Discussion of results 8Conclusions 14References cited 15

FIGURES

1-3. Box plots showing:1. Results for a series of solutions based on a National Institute of Standards

and Technology standard reference material preserved with acid dichromateor hydrochloric acid 9

2. Results for U.S. Geological Survey standard reference water samples Hg24and Hg26 and two dilutions of a Baker Instra-Analyzed standard preservedin acid dichromate 11

3. The effects of using argon instead of nitrogen as the purge gas in coldvapor–atomic fluorescence spectrometry 12

4. Graph showing linear regression analysis of results for sample solutionsbased on two reference standards (experimental) and National Instituteof Standards and Technology standard reference material (theoretical) 13

TABLES

1. Method detection limit and analytical precision for the determination ofmercury by cold vapor–atomic fluorescence spectrometry 3

2. Typical instrument operating conditions 73. Average percent spike recoveries for mercury in reagent-water, filtered and

unfiltered surface-water, and filtered and unfiltered ground-water matrices 14

ContentsIV

CONVERSION FACTORSMultiply By To obtain

Lengthmicrometer (µm) 3.94 x 10-5 inchmillimeter (mm) 3.94 x 10-2 inchnanometer (nm) 3.94 x 10-8 inch

Massgram (g) 3.53 x 10-2 ounce, avoirdupoismilligram (mg) 3.53 x 10-5 ounce, avoirdupoisnanogram (ng) 3.53 x 10-11 ounce, avoirdupois

Volumeliter (L) 2.64 x 10-1 gallonmicroliter (µL) 2.64 x 10-7 gallonmilliliter (mL) 2.64 x 10-4 gallon

Pressurekilopascal (kPa) 1.45 x 10-1 pounds per square inch

Degrees Celsius (oC) may be converted to degrees Fahrenheit (oF) by using the followingequation:

oF = 9/5 (oC) + 32.

ABBREVIATED WATER-QUALITY UNITS

mg/L milligrams per literµg/L micrograms per literµg/mL microgram per milliliterng/L nanograms per litermL/min milliliters per minuteL/min liters per minute

Contents V

ABBREVIATIONS AND ACRONYMS

ASTM American Society for Testing and MaterialsCASRN Chemical Abstract Service Registry NumberCV–AAS cold vapor–atomic absorption spectrometryCV–AFS cold vapor–atomic fluorescence spectrometryHCl hydrochloric acidM Molar (moles per liter)LT–MDL long-term method detection levelMDL(s) method detection limit(s)MRL method reporting levelHNO3 nitric acidN Normal (acid equivalents per liter)NIST National Institute of Standards and TechnologyNWQL National Water Quality Laboratorys secondsp. gr. Specific gravitySRM standard reference materialSRWS(s) U.S. Geological Survey standard reference water sample(s)USEPA U.S. Environmental Protection AgencyUSGS U.S. Geological Survey< less than± plus or minus

GLOSSARY

MDL — The method detection limit (MDL) is defined as the minimum concentration of anelement that can be measured and reported with 99-percent confidence that the elementconcentration is greater than zero and is determined from analysis of a sample in a given matrixthat contains the element of interest (U.S. Environmental Protection Agency, 2000).

Introduction 1

Methods of Analysis by the U.S. Geological Survey NationalWater Quality Laboratory—Determination of Organic Plus InorganicMercury in Filtered and Unfiltered Natural Water with Cold Vapor –Atomic Fluorescence Spectrometry

By John R. Garbarino and Donna L. Damrau

Abstract

An analytical method using cold vapor–atomic fluorescence spectrometry wasdeveloped by the U.S. Geological Survey in2001 for the determination of organic plusinorganic mercury in filtered and unfilterednatural water. This method was developed toeliminate the use of acid dichromatepreservative and to provide capability tomeasure ambient mercury concentrations innatural water. Dissolved mercury includes alloxidizable mercury species present in naturalwater that has been filtered through a 0.45-micrometer pore size capsule filter. Whole-water recoverable mercury includes dissolvedmercury species and mercury speciesadsorbed to particulate matter in unfilterednatural water. Mercury species can includeelemental mercury, mercury (II), mercury (II)complexes, various alkyl- and phenyl-mercury compounds, and other forms ofmercury. In this method, samples arecollected and processed according to standardU.S. Geological Survey protocols. Samplesare preserved onsite with 6N hydrochloricacid in a ratio of 1 to 100 in a borosilicate-glass bottle with fluoropolymer-lined cap.Mercury species are oxidized to mercury (II)by using bromine monochloride; excessoxidation reagent is neutralized withhydroxylamine hydrochloride. Elementalmercury produced after adding stannouschloride is purged from the solution withultrapure argon gas into a cell in which themercury concentration is measured by atomicfluorescence emission at 253.7 nanometers.The analytical response is linear up to 125

nanograms per liter (ng/L) of mercury, andthe short-term method detection limit is about5 ng/L. The analytical variability at 50 ng/Lis about 10 percent.

This report describes the method andcompares the use of hydrochloric acid to aciddichromate as a field preservative. Ambientmercury concentrations in hydrochloric acid-preserved samples stored in borosilicate-glassbottles with fluoropolymer-lined caps areshown to be stable for at least 30 days.Mercury concentrations are stable for at least5 months after bromine monochloride isadded to the sample bottles in the laboratory.The long-term average percent recoveries at20, 45, and 75 ng/L in reagent water, filteredand unfiltered ground water, and filtered andunfiltered surface water range from 89 to 108,96 to 103, and 94 to 98 percent, respectively.

INTRODUCTION

The ambient concentration for dissolvedand whole-water recoverable mercury isgenerally at the parts-per-trillion level(nanograms-per-liter, ng/L) for natural-watersamples. Measurement at this concentrationlevel is important because mercurybioaccumulates in living organisms. Theprevious method (Fishman and Friedman,1989, p. 289–291) that used cold vapor–atomic absorption spectrometry (CV–AAS)was not capable of measuring such lowconcentrations. Cold vapor–atomicfluorescence spectrometry (CV–AFS) iscapable of measuring 5 ng/L mercury directly,without sample preconcentration. By

Determination of Organic Plus Inorganic Mercury in Filtered and UnfilteredNatural Water with Cold Vapor-Atomic Fluorescence Spectrometry

2

eliminating sample preconcentration, analysisthroughput is increased while maintaining thecapability to determine mercuryconcentrations at levels typically regulated bywater-quality standards. The mercuryconcentration, organic plus inorganic species,is determined in filtered and unfilterednatural-water samples by using a brominatingdigestion procedure that minimizesinterferences. Samples are preserved withhydrochloric acid instead of acid dichromatesolution, which has previously been used.This change was necessary because shipmentof samples preserved with dichromatesolution is restricted, and the cost of disposingof chromium waste is expensive. In addition,hydrochloric acid sample preservation hasbeen shown as effective as samplepreservation with acid dichromate solution.

This report describes a method fordetermining mercury in natural-water

samples. It was developed by the U.S.Geological Survey (USGS) for use in theNational Water Quality Laboratory (NWQL).It is rapid, more efficient, and can detectlower concentration levels compared toprevious USGS methods (Fishman andFriedman, 1989, p. 289–291). This methodsupplements other methods of the USGS forthe determination of mercury that aredescribed by Fishman and Friedman (1989)and by Fishman (1993). The new method wasimplemented at NWQL in April 2001.

This report provides a detaileddescription of all aspects of the method,including the instrumentation, the reagents,the analytical procedure, and the quality-control procedures. Mercury concentration isreported by NWQL in micrograms per literinstead of nanograms per liter, theconcentration unit used throughout this report,because of current (2001) limitations in thedata base.

ANALYTICAL METHOD

Inorganic Constituents and Method, Lab, and Parameter Codes

Constituent MethodLabcode

Parameter andmethod code

Mercury, water filtered, organic + inorganic, g/L I-2464-01 2707 71890CMercury, water unfiltered, organic + inorganic,g/L I-4464-01 2708 71900D

1. Application

This method is used to determineconcentrations of organic plus inorganicmercury in filtered and unfiltered natural-water samples. Bromine monochloride isused to oxidize elemental mercury, mercurycomplexes, various alkyl- and phenyl-mercury compounds, and other forms ofmercury. The upper limit of the linear

concentration range is 125 ng/L mercury.The short-term method detection limit (MDL,see Glossary) is 5 ng/L mercury (table 1).

2. Summary of Method

Concentrations of organic plus inorganicmercury in filtered and unfiltered natural-watersamples are determined by using CV–AFS.The method is based on U.S. Environmental

Analytical Method 3

Protection Agency (1999) Method 1631 butdoes not use gold amalgamation for mercurypreconcentration. By eliminating thepreconcentration procedure, the MDL is atleast a factor of 10 greater, yet the methodprovides the determination of mercury atambient concentrations. Mercury species in anatural-water sample are oxidized to mercury(II) inside the borosilicate-glass sample bottleby using bromine monochloride; excessoxidant is neutralized with hydroxylaminehydrochloride. Mercury (II) is reduced toelemental mercury by using stannouschloride, and is purged from the solution withultrapure argon gas into a cell in which themercury concentration is measured by atomicfluorescence emission at 253.7 nanometers(nm).

Table 1. Method detection limit and analyticalprecision for the determination of mercury by coldvapor–atomic fluorescence spectrometry

[ng/L, nanograms per liter; MDL, method detection limit]

ElementMean,ng/L

Standarddeviation,

ng/L

t-statistic

Degreesof

freedom

MDL,ng/L

Mercury 11 2 2.492 24 5

3. Sample Collection and Storage

Sample-collection protocol must followeither the procedure outlined for traceelements in Horowitz and others (1994) or formercury in Olson and DeWild (1999). Thelevel of contamination related to fieldcollection and processing for either protocol,however, must be established by using anequipment blank. Filtered and unfilterednatural-water samples are preserved withhydrochloric acid to a pH less than 2.Samples are stored in clean borosilicate-glasssample bottles with fluoropolymer-lined caps.Bottles are prepared by rinsing once with

10 percent (by volume) mercury-free nitricacid and three times with deionized waterbefore air-drying on a laminar-flow cleanbench. The sample bottle must be rinsed witha portion of the filtered or unfiltered naturalwater prior to filling the bottle with about 200mL of sample. Every sample must bepreserved with 2 mL of 6N hydrochloric acid(HCl); more acid might be required to lowerthe pH to 2 in some matrices. It is importantthat field personnel use 6N HCl prepackagedin polypropylene vials that can be obtainedfrom the USGS Ocala Water Quality andResearch Laboratory in Florida (part number06910) because the quality control iscontinually monitored. Ambient mercuryconcentrations are stable for 1 month inpreserved samples. Nevertheless, samplesshould be shipped to the laboratory foranalysis as quickly as possible.

4. Quality-Control Samples, SamplingMethods, Sample-CollectionEquipment, and Cleaning Procedures

4.1 Quality-control samples. Collectionof quality-control (QC) samples is a requiredcomponent of sample collection for water-quality studies. QC samples are collected,usually onsite, to identify, quantify, anddocument bias and variability in data thatresult from collecting, processing, shipping,and handling of samples by field andlaboratory personnel. The type, number, anddistribution of QC samples are determined bythe design and data-quality requirements ofthe study. Detailed discussion of the typesand purposes of quality-control samples isprovided in Wilde and others (1999).

The primary purpose of a blank sample isto identify potential sources of samplecontamination and to assess the magnitude ofmercury contamination. Field blanks arecollected and processed onsite in the same


4

manner and by using the same equipment asfor the environmental samples. The fieldblank is an aliquot of blank water that isprocessed sequentially through eachcomponent of the sampling system. Thesource water needed for blank samples mustbe produced and certified by a laboratory tohave mercury concentrations that do notexceed a specified method detection limit.Inorganic-free blank water must be used forall equipment field blanks (Horowitz andothers, 1994, p. 22).

Replicate samples also may becollected. The primary purpose of replicatesamples is to quantify the variability in all orpart of the sampling and analysis process.Replicates, environmental samples collectedin duplicate, triplicate, or higher multiples, areconsidered identical in composition and areanalyzed for the same chemical constituents.

4.2 Sampling methods. Use methodsthat allow collection of water samples thataccurately represent the water-qualitycharacteristics of the surface water or groundwater at a given time or location. Detaileddescriptions of sampling methods used by theUSGS for obtaining surface-water samples,ground-water samples, and sample-processingprocedures (splitting, filtration, and shipping)are provided by Wilde and others (1999).

4.3 Sample-collection equipment. Usesample-collection equipment and automaticsamplers that are free of tubing, gaskets, andother components made of metal or that mightsorb mercury from the water. The bestmaterial for sample-collection and processingequipment is any type of fluorocarbonpolymer.

4.4 Cleaning procedures. Follow theprocedures outlined in either Horowitz andothers (1994) or Olson and DeWild (1999).

5. Contamination and Interferences

Contamination must be avoided becauseof the inherently low mercury concentrationfound in natural water. The analyst and fieldpersonnel need to be attentive to potentialsources of contamination. Samplingequipment must be cleaned, and samples needto be collected as described by Horowitz andothers (1994) or by Olson and DeWild(1999). Every sample must be preserved with2 mL of 6N HCl that can be obtained from theUSGS Ocala Water Quality and ResearchLaboratory in Florida. Samples mightbecome contaminated during analysis bycarryover from a previous sample that hadunusually high mercury concentration.Contamination from airborne sources can beminimized by placing the autosampler insidea plastic enclosure pressurized with mercury-free nitrogen.

Gold, silver, and iodide interfere withthe determination of mercury by this method.Elemental mercury is amalgamated by gold orsilver. The recovery of 2.5 ng/L of mercuryin the presence of 5 to 100 mg/L iodideranges from 100 to 0 percent (U.S.Environmental Protection Agency, 1996;Bloom, 1995). After analyzing a sample thatcontains greater than 30 mg/L iodide, it maybe necessary to clean the analytical systemwith 4N HCl (U.S. Environmental ProtectionAgency, 1999; Bloom, 1995). Most forms ofmercury are oxidized by the brominemonochloride solution; however, the recoveryof mercury bound within microbial cellsmight require additional ultravioletphotooxidation (U.S. EnvironmentalProtection Agency, 1999). Buildup ofcondensation in the vapor generator orfluorescence cell will degrade the analyticalsignal. Condensation can be eliminated bywrapping heating tape around the vaporgenerator and by using a drying tube in front

Analytical Method 5

of the fluorescence cell. The drying tubeshould be replaced after about 1 year ofoperation. Ultrapure argon (99.998 percent)must be used for the purge gas to minimizethe possibility of fluorescent quenching byimpurities.

6. Apparatus and Instrumentation

6.1 Labware. Use clean Type A glassvolumetric flasks to prepare all solutions.Store mercury stock solutions in borosilicate-glass bottles with fluoropolymer-lined caps.The accuracy of all pipets and volumetricflasks should be regularly verifiedgravimetrically or by using an automaticvolume-calibrating spectrophotometer systembefore preparing standard solutions.

6.2 Instrumentation. Severalcommercial automated CV–AFS systems areavailable. These systems are composed of aspectrometer, mercury-vapor generator,autosampler, and computer. The majorcomponents of the spectrometer are a mercuryline source, a quartz flow-throughfluorescence cell, and a photometer. Themercury-vapor generator is composed ofseveral peristaltic pumps and a series ofcomputer controlled, time-actuated solenoidvalves that introduce sample and reagents, agas-liquid separator, and a membrane dryertube that removes moisture. The autosamplerand computer automate sample introduction,data acquisition, and quantitation.

7. Reagents and CalibrationStandards

All solutions used in this method must beverified to have mercury-contaminantconcentration, after the prescribed dilution, thatis less than the MDL. ASTM Type I reagentwater (American Society for Testing andMaterials, 2000) must always be used to

prepare solutions. Heating each compoundused in this method, with the exception of thehydrated stannous chloride, in a mufflefurnace at 250C for 8 hours will volatilizemercury impurities. All solutions must bestored in designated borosilicate-glass bottleswith fluoropolymer-lined caps. Allcalibration standards are stable for 1 month.

7.1 Argon gas is used as the purge(carrier) gas. Its purity must be at least99.998 percent.

7.2 Nitrogen gas is used as the sheathgas. Its purity must be at least 99.998percent.

7.3 Hydroxylamine hydrochloride(NH2OH·HCl) Chemical Abstract ServiceRegistry Number (CASRN) 5470-11-1.

7.4 Potassium bromate (KBrO3)CASRN 7758-01-2.

7.5 Potassium bromide (KBr) CASRN7758-02-3.

7.6 Stannous chloride (SnCl2·2H2O)CASRN 10025-69-1.

7.7 Nitric acid solution, 0.4 percent (byvolume). Add 4 mL of concentrated nitricacid [HNO3, 16 M, specific gravity (sp. gr.)1.41] to 500 mL of water in a 1-L volumetricflask. Bring to volume with water. Thissolution is used to rinse between sampleanalyses.

7.8 Hydroxylamine hydrochloride(NH2OH·HCl) solution, 15 percent (byweight). Dissolve 30 g of NH2OH·HCl in a200-mL volumetric flask that contains about10 mL of water. Bring to volume with water.Solution must be prepared weekly.


6

7.9 Bromine monochloride (BrCl)solution. This solution is used to oxidizemercury species in a sample to mercury (II).Dissolve 10.8 g of KBr in 1 L of concentratedhydrochloric acid (HCl, 12 M, sp. gr. 1.19) ina borosilicate-glass bottle. Place a cleanmagnetic stir bar in the bottle and stir forabout 1 hour in the fume hood. Slowly add15.2 g KBrO3 to the solution while stirring.(NOTE: This process produces copiousquantities of free halogens, such as Cl2, Br2,BrCl.) When all of the KBrO3 has beenadded, the color of the solution should changefrom yellow to red to orange. Loosely cap thebottle, and stir another hour before tightening.(Paragraph modified: November 14, 2002 jrg)

7.10 Stannous chloride (SnCl2)solution, 2 percent (weight-to-volume ratio).This solution is used to reduce mercury (II) toelemental mercury. Dissolve 20 g ofSnCl2·2H2O into a 1-L volumetric flask thatcontains 200 mL of water. Slowly add 30 mLconcentrated HCl, mix, and bring to volumewith water. Allow solution to equilibrate 1hour before use. Solution must be prepareddaily.

7.11 Primary mercury stock solution,10,000 mg/L. National Institute of Standardsand Technology (NIST) standard referencematerial (SRM) 3133. This reference materialis stable until the NIST expiration date.

7.12 Secondary mercury stock solution,1.00 mg/L. Pipet 100L of primary standardsolution into a 1-L volumetric flask, and 5 mLof BrCl solution, and bring to volume withwater. Solution is stable for 1 year.

7.13 Tertiary mercury stock solution,2.0 g/L. Pipet 2.0 mL of secondary stocksolution into a 1-L volumetric flask, and 5 mLof BrCl solution, and bring to volume withwater. Solution is stable for 1 month.

7.14 Calibration standard 1, 0 ng/L. Add1 mL of BrCl solution to a 200-mL volumetricflask and bring to volume with water.

7.15 Calibration standard 2, 5.0 ng/L.Add 0.5 mL of the tertiary mercury stocksolution and 1 mL of BrCl solution to a 200-mL volumetric flask and bring to volume withwater.




7.19 Calibration standard 6, 100.0ng/L. Add 10.0 mL of the tertiary mercurystock solution and 1 mL of BrCl solution to a200-mL volumetric flask and bring to volumewith water.

8. Sample Preparation

Samples and calibration solutions areprocessed identically. Mercury species areoxidized to mercury (II) by adding 1 mL ofBrCl solution (see section 7.9) to 200 mL offield-acidified, filtered or unfiltered, natural-water sample. The sample is digested for atleast 12 hours at room temperature prior to

Analytical Method 7

analysis. If the yellow color disappears, addmore BrCl solution to fully digest the sample.The sample is completely digested when theyellow color persists for at least 12 hours.Record the total volume of BrCl solutionadded so that the final mercury concentrationcan be adjusted for dilution. Mercury boundwithin microbial cells also might requireultraviolet photooxidation (U.S.Environmental Protection Agency, 1999).Some highly organic matrices might requirelonger oxidation times at an elevatedtemperature. The sample bottles can beplaced in an oven set at 50C for 6 hourswhen elevated temperature is required.Nevertheless, samples should be analyzed asquickly as possible after the oxidationprocedure.

Just prior to analysis, 10 mL of BrCl-oxidized sample is pipetted into anautosampler borosilicate-glass test tube.Adding 40 L of hydroxylaminehydrochloride solution (see section 7.8)neutralizes excess BrCl. Allow about 5minutes for the reaction to reach completion.The disappearance of the yellow colorindicates that all the BrCl has been destroyedand no traces of halogens remain.

9. Analytical Procedure

The analytical procedure for the methodis fully automated. After calibrationstandards and unknown samples have beenprocessed by using the procedure outlined insection 8, they are loaded into theautosampler. The system introduces a givenvolume of sample and SnCl2 solution (seesection 7.10) with a series of peristalticpumps and solenoid valves. The mercury (II)present is reduced to elemental mercury thatis purged from the reactor into thefluorescence cell with argon gas. Themercury concentration is determined byrelating the signal peak area for the samplesto the peak area for the calibration standards.

Refer to the NWQL Standard OperatingProcedure IM0348.0 (D.L. Damrau, U.S.Geological Survey, written commun., 2001),Zellweger Analytics Inc. (1996; 1997), andWendt (1997) for details of the operation andmaintenance of the instrumentation used inthis method. Typical operating conditions arelisted in table 2.

Table 2. Typical instrument operating conditions

[kPa, kilopascal; lbs/in2, pounds per square inch; mL/min,milliliters per minute; L/min, liters per minute; s, second]

Parameter Setting

Mercury-vapor generator flow rates

300 mL/min

2.75 L/min

Argon at 276 kPa (40 lbs/in2)Purge (carrier) gas - - - - - - - - - - - - -Nitrogen at 276 kPa (40 lbs/in2)Dryer gas - - - - - - - - - - - - - - - - - - -Sheath gas - - - - - - - - - - - - - - - - - - 250 mL/min

Mercury-vapor generator timing

18 s170 s

Sample delay - - - - - - - - - - - - - - - - - -Wash period - - - - - - - - - - - - - - - - - -Load period - - - - - - - - - - - - - - - - - - - 45 s

Spectrometer1,0006.50Ratio

Calibration range- - - - - - - - - - - - - - - -Fine gain - - - - - - - - - - - - - - -- - - - - -Mode switch - - - - - - - - - - - - - - - - - -Integration time - - - - - - - - - - - - - - - - 0.25 s

10. Calculations

No calculations are required exceptwhen samples are diluted and dilution factorsare applied.

11. Reporting Results

Mercury concentration is reported byNWQL in micrograms per liter (g/L) insteadof nanograms per liter (ng/L) because of thecurrent (2001) limitations of the data base.The number of significant figures reported arelisted as follows:

If the mercury concentration is lessthan the long-term methoddetection level (LT–MDL), then the


8

result is reported as less than (<)the method reporting level (MRL),in micrograms per liter. Samplesbetween the LT–MDL and MRLwill be “E” coded to indicate anestimated concentration.

If the mercury concentration isgreater than or equal to the MDL,but less than 0.100 g/L, then theresult is reported to the nearest0.001 g/L.

If the mercury concentration isgreater than 0.100 g/L, then theresult is reported to the nearest0.01 g/L.

DISCUSSION OF RESULTS

The CV–AFS mercury method wasdeveloped to eliminate the use of the aciddichromate preservative and to provide amethod that is capable of measuring ambientmercury concentrations in natural water. AU.S. Department of Transportation regulationon the shipment of natural-water samplespreserved with acid dichromate requirespackaging and shipment protocols that areexpensive. In addition, it is expensive todispose of samples that contain highconcentrations of chromium because it is apriority pollutant. Several U.S.Environmental Protection Agency (USEPA)methods [for example, Method 1631 (U.S.Environmental Protection Agency, 1999], usehydrochloric acid (HCl) as a preservative. Awater sample preserved with HCl passes allrequired corrosion tests and can be shippedwithout expensive packaging.

A filtered or unfiltered natural-watersample is stored in a 250-mL borosilicate-glass bottle with a fluoropolymer-lined capafter adding 2 mL of 6N HCl to about 200 mLwater (water level at the shoulder of thebottle). More acid must be added to samples

that have high alkalinity to ensure that the pHis less than 2. If more than 2 mL of HCl isused, the total volume of acid should be listedon the Analytical Services Request form sothat adjustments for dilution can be applied.It is important that samples are preserved withHCl specifically packaged and tested for suchuse. Polypropylene vials that contain HClpreservative can be obtained from the USGSOcala Water Quality and Research Laboratory(part number 06910). The HCl preserves themercury concentration in water samples for atleast 30 days; the relative standard deviationof the results was less than 10 percent for asolution that had 10 ng/L mercury and thatwas analyzed two to three times a week for 1month. No significant difference was foundbetween the HCl and the previously used(prior to April 2001) acid dichromatepreservatives. The average concentration fora sample that had 10 ng/L mercury preservedwith HCl and that was analyzed repetitivelyfor 90 days was not significantly different atthe 95-percent confidence level (p=0.20) thanthe average concentration for a similarsolution preserved with acid dichromate.Results obtained for solutions that had 5, 10,25, 50, and 100 ng/L mercury (fig. 1) showedthat the variation, as a function of mercuryconcentration, was similar for bothpreservation agents.

Mercury species in natural water cantake various forms, including elementalmercury, mercury complexes, plus variousalkyl- and phenyl-mercury compounds.Oxidation of such species to mercury (II) byusing BrCl makes it possible to determineconcentrations of organic plus inorganicmercury and also serves as a preservative toincrease sample-holding times. Experimentshave indicated that samples are stable for upto 3 months after the addition of BrCl. Thepercent-relative standard deviation for asample that had 10 ng/L mercury that wasanalyzed more than 24 times for 3 months

Discussion of Results 9

Figure 1. Results for a series of solutions based on a National Institute of Standards and Technologystandard reference material preserved with acid dichromate or hydrochloric acid (HCl). Standardreference material concentration is in nanograms per liter.

5 10 25 50 100

0

20

40

60

80

100

120

140

HCl

EXP

ER

IME

NTA

LM

ER

CU

RY

CO

NC

ENT

RA

TIO

N,

INN

AN

OG

RA

MS

PE

RLI

TER

MERCURY STANDARD REFERENCE MATERIAL

0

20

40

60

80

100

120

140

Acid dichromate

EXPLANATION

Percentile90th

75th

50th

25th

10th

= 1st and 99th percentile= Mean


10

was about 15 percent, an acceptable level ofvariability considering the concentration leveland daily deviations in instrumentperformance and calibration standards.

The bias and variability of the methodwas tested by using standard referencematerials from several different sources. Themethod performance is validated through therepetitive analysis of reagent-water, ground-water, and surface-water matrices spiked atthree different mercury concentrations.Comparison of the analytical performancebetween the new CV–AFS method and thepreviously used CV–AAS (prior to April2001) is not feasible because the methodreporting level for the CV–AAS is only about100 ng/L. This reporting level is about 20times greater than the MDL for the new CV–AFS method, and 100 ng/L corresponds to theupper calibration limit. Onsite samplescollected before April 2001 were preservedwith acid dichromate, thus making directcomparisons problematic. Furthermore,during a typical year, more than 90 percent ofthe results for samples analyzed by CV–AASfor dissolved and whole-water recoverablemercury were less than or equal to 100 ng/L.

The accuracy of CV–AFS was tested byusing two USGS standard reference watersamples (SRWS). Calibration standards wereprepared from NIST SRM-3133 in aciddichromate because all available referencewater samples were preserved in the samepreservative. The SRWSs were diluted by afactor of 10 because the most probable meanmercury concentration for all availableSRWSs exceeded the calibration range.Additional mercury solutions preserved inacid dichromate were prepared at 10 and 50ng/L from Baker Instra-Analyzed 1,000-g/mL standard solution to provide anotherreference check. These four referencesolutions were analyzed repetitively for 2months after calibrating the instrument with

NIST standards preserved with aciddichromate. Results for the diluted SRWSscompared favorably with expectedconcentrations (fig. 2). The experimentalresults of 38±2 (n=21) and 71±3 ng/L (n=19)correspond to the theoretical means of 420and 700 ng/L, respectively, after beingadjusted for dilution. The results for the 10-and 50-ng/L solutions prepared from theBaker Instra-Analyzed standard wereacceptable at 10±2 and 55±3 ng/L,respectively.

Either nitrogen or argon gas has beenused in analytical methods for mercury topurge elemental mercury from the reactorcell. Experimental results (fig. 3) show thatthere is little difference between the two gaseswhen mercury concentrations exceed 10 ng/L.The use of argon gas significantly increasesthe precision at 10 ng/L, because thesensitivity of the measurement is increased byabout a factor of 2 over nitrogen gas. Thus,the use of argon gas is necessary formeasuring ambient mercury concentrations.The accuracy and variability of using argonwas established by analyzing referencestandards prepared from two different sourcesrepetitively for 1 month. Mercury solutionspreserved in HCl were prepared from BakerInstra-Analyzed 1,000-g/mL (25 and 75ng/L) and Spex CertiPrep 10-g/mL (10 and50 ng/L) standard solutions. Experimentalconcentrations were determined for thesesolutions relative to NIST calibrationstandards. Linear regression analysis (fig. 4)of results shows a significant correlationbetween the experimental (Baker and Spex)solutions and the theoretical concentrations(slope=0.925, correlation coefficient=0.9992).The variability ranged from 5 to 15 percentand was a function of concentration. Inaddition, the stability of the preservation andinstrumentation is supported by the data thatwere accumulated for a month.


Figure 2. Results for U.S. Geological Survey standard reference water samples Hg24 and Hg26 andtwo dilutions of a Baker Instra-Analyzed standard preserved in acid dichromate. Mercuryconcentrations for standard reference materials are in nanograms per liter (ng/L).

EXPLANATIONPercentile

90th

75th

50th

25th

10th


Hg24 Hg26 10 50

0

20

40

60

80

100

1:10 DilutionMean=420 ng/L

Baker

Baker1:10 DilutionMean=700 ng/L

EX

PE

RIM

EN

TA

LM

ER

CU

RY

CO

NC

EN

TRA

TIO

N,

INN

AN

OG

RA

MS

PE

RL

ITE

R

MERCURY STANDARD REFERENCE MATERIAL


12

Figure 3. The effects of using argon instead of nitrogen as the purge gas in cold vapor–atomicfluorescence spectrometry.

0

20

40

60

80

100

120

140Nitrogen

5 10 25 50 100

0

20

40

60

80

100

120

140Argon

EX

PE

RIM

EN

TA

LM

ER

CU

RY

CO

NC

EN

TR

AT

ION

,IN

NA

NO

GR

AM

SP

ER

LIT

ER

THEORETICAL MERCURY CONCENTRATION,IN NANOGRAMS PER LITER

EXPLANATION

Percentile90th

75th

50th

25th

10th



Figure 4. Linear regression analysis of results for sample solutions based on two referencestandards (experimental) and National Institute of Standards and Technology standard referencematerial (theoretical). Error bars correspond to the standard deviation.

0 20 40 60 800

20

40

60

80

Slope: 0.925Y-intercept: 3.3Correlation coeff icient: 0.9992

EXP

ER

IME

NT

AL

ME

RC

UR

YC

ON

CE

NT

RA

TIO

N,

INN

AN

OG

RA

MS

PE

RL

ITE

R

THEO RETICAL MERCURY CONCENTRATION,IN NANOGRAMS PER LITER


14

The percentage recovery of results forvarious concentrations of mercury spiked intoreagent water, filtered and unfiltered surfacewater, and filtered and unfiltered groundwater was used to validate the method. Datawere acquired during 7 nonconsecutive days.No more than two sets of data were acquiredin 1 day. The range of average percentrecovery for all filtered and unfilteredmatrices at 20, 45, and 75 ng/L was 89 to108,96 to 103, and 94 to 98 percent, respectively(table 3). The variability in the recovery of

20 ng/L, a concentration about four times theshort-term MDL, was about twice that ofhigher spike concentrations. In addition, thevariability in recoveries at 20 ng/L for theunfiltered matrices was about twice those forcorresponding filtered matrices. There waslittle difference, however, in variability forfiltered and unfiltered matrices for spikeconcentrations of 45 and 75 ng/L. Sorption toparticulate matter might contribute to theincrease in variability for the 20-ng/L-mercury spike in unfiltered matrices.

Table 3. Average percent spike recoveries for mercury in reagent-water, filtered and unfiltered surface-water, and filtered and unfiltered ground-water matrices

[Average is based on 9 to 11 replicates acquired on 7 nonconsecutive days; ng/L, nanograms per liter; %, percent;±, plus or minus; filtered, surface or ground water processed through a 0.45-micrometer membrane]

Mercury spike concentration, ng/L20 45 75Matrix

Averagerecovery, %

Averagerecovery, %

Averagerecovery, %

Reagent water - - - - - - - - - - - - - - - - 94±11 100±6 98±8

Surface water, filtered - - - - - - - - - - - 89±8 100±9 96±6

Surface water, unfiltered - - - - - - - - - 108±20 101±10 95±8

Ground water, filtered - - - - - - - - - - - 91±13 96±6 95±9

Ground water, unfiltered - - - - - - - - - 102±20 103±10 94±7

CONCLUSIONS

The new cold vapor–atomicfluorescence spectrometric (CV–AFS)method for the determination of inorganicplus organic mercury species in filtered andunfiltered natural water has severaladvantages when compared to the previouslyused (prior to April 2001) cold vapor–atomicabsorption spectrometric method. Primaryadvantages are improved method detectionlimit and elimination of a corrosivepreservative. The CV–AFS method has ashort-term method detection limit of 5 ng/Lwithout using gold amalgamation for mercury

preconcentration. Without mercurypreconcentration, analysis throughput isincreased while maintaining the capability ofdetermining mercury at concentrationstypically regulated by water-quality standards.Mercury samples are preserved withhydrochloric acid (HCl) instead of aciddichromate. Although acid dichromate hasbeen used extensively to preserve mercurysamples, shipping and disposal regulationshave made continued use too expensive.Experiments suggest that there is nosignificant difference, however, between the

References Cited 15

average mercury concentration in a watersample preserved with HCl and one preservedwith acid dichromate. Ambient mercuryconcentrations in samples preserved with HClin borosilicate-glass bottles with fluoro-polymer-lined caps have been shown to bestable for at least 30 days. Oxidation ofinorganic and organic mercury species withbromine monochloride makes it possible todetermine total recoverable mercury andextends the sample-holding time to 3 months.

The bias and variability of the methodwere established using standard referencematerials and spike recovery results.Comparison of the analytical performancebetween CV–AFS and the previously usedmethod were not feasible because of thesizable disparity in method detection limits.The concentration of the upper calibrationstandard used by the new method correspondsto the detection limit of the old method.Results for USGS standard reference watersamples were acceptable with respect to themost probable concentration. Other solutionsthat have varying mercury concentrationswere prepared from certified standards fromtwo different sources. Linear regression ofthe results for these solutions weresignificantly correlated (correlationcoefficient=0.9992) with standard solutionstraceable to National Institute of Standardsand Technology materials. The analyticalvariability is expected to extend from 15 to 5percent with increasing concentrationbounded by the calibration range. Theaverage percent recovery of 20, 45, and 75ng/L of mercury in reagent-water, surface-water, and ground-water matrices wasdetermined during 7 nonconsecutive days.The average percent recoveries for all filteredand unfiltered matrices ranged from 89 to108(20 ng/L), 96 to 103 (45 ng/L), and 94 to 98percent (75 ng/L). The variability for therecovery of the 20-ng/L spike, a concentrationabout four times the method detection limit,was about twice that of higher spike

concentrations. In addition, the variability inpercentage recoveries for unfiltered matriceswas twice that for matching filtered matrices,possibly because of sorption to particulatematter.

Analytical results from the new CV–AFS method will most likely impactinterpretation of long-term trends observed inwater-quality studies involving mercury. Themethod detection limit of 5 ng/L is about afactor of 20 times lower than the previouslyused method. More than 90 percent of themercury concentrations reported with theprevious analytical method during a typicalyear were less than or equal to 100 ng/L priorto April 2001. Implementation of the CV–AFS method will likely provide moredefinitive results because its analytical rangeextends well below 100 ng/L.

REFERENCES CITED

American Society for Testing and Materials,2000, Annual Book of ASTMStandards, Section 11, Water:Philadelphia, American Society forTesting and Materials, v. 11.01,D1193, p. 10.

Bloom, N.S., 1995, Trace metals and ultra-clean sample handling: EnvironmentalLaboratory, v. 7, p. 20.

Fishman, M.J., 1993, Methods of analysis bythe U.S. Geological Survey NationalWater Quality Laboratory—Determination of inorganic andorganic constituents in water andfluvial sediments: U.S. GeologicalSurvey Open-File Report 93-125,217 p.

Fishman, M.J., and Friedman, L.C., eds., 1989,Methods for determination of inorganicsubstances in water and fluvialsediments: U.S. Geological SurveyTechniques of Water-ResourcesInvestigations, Book 5, Chap. A1, p.289–291.


16

Horowitz, A.J., Demas, C.R., Fitzgerald,K.K., Miller, T.L., and Rickert, D.A.,1994, U.S. Geological Surveyprotocol for the collection andprocessing of surface-water samplesfor the subsequent determination ofinorganic constituents in naturalwater: U.S. Geological SurveyOpen-File Report 94-539, 57 p.

Olson, M.L., and DeWild, J.F., 1999, Low-level collection techniques andspecies-specific analytical method formercury in water, sediment, andbiota: U.S. Geological SurveyWater-Resources InvestigationsReport 99-4018b, 11 p.

U.S. Environmental Protection Agency, 1996,Method 245.7—Determination ofultra-trace level (ng Hg/L) totalmercury in water by cold vaporatomic fluorescence spectrometry:U.S. Environmental ProtectionAgency, Revision 1.1, NationalExposure Research Laboratory,Research Triangle Park, Office ofResearch and Development, 29 p.

_______1999, Method 1631—Revision B:Mercury in water by oxidation, purgeand trap, and cold vapor atomicfluorescence spectrometry: EPA821-R-99-005, May 1999, 33 p.

_______2000, Guideline establishing testprocedures for the analysis ofpollutants (Part 136, Appendix B.Definition and procedure for thedetermination of the methoddetection limit—Revision 1.11):U.S. Code of Federal Regulations,Title 40, July 1, 2000, p. 310–313.

Wendt, Karin, 1997, QuikChem Method 10-138-37-1-A, Determination of mercuryby cold vapor generation and atomicfluorescence detection: Milwaukee,Wisconsin, LaChat Instruments, 20 p.

Wilde, F.D., Radtke, D.B., Gibs,Jacob, and Iwatsubo, R.T., eds.,1999, National field manual for thecollection of water-quality data: U.S.Geological Survey Techniques ofWater-Resources Investigations,Book 9, Chaps. A1 through A6.

Zellweger Analytics Inc., 1996, QuikChemmercury analyzer system operationmanual: Milwaukee, Wisconsin,April 12, 1996, 66 p.

_______1997, QuikChem 8000 automated ionanalyzer Omnion FIA SoftwareManual: Milwaukee, Wisconsin,April 22, 1997, 64 p.

Documents

Methods of Analysis by The U.S. Geological Survey National ...Inorganic Constituents and Method, Lab, and Parameter Codes Constituent Method Lab code Parameter and method code Mercury,