PBS4-178425
Multielement# 1 Analytical Techniques for
Hazardous Waste Analysis
The Stata-of-ths-Art
Montgomery (Jaaes M.), Inc., Pasadena, CA
Prepared for
Bnvir omaer. til Monitoring Systeas Lab.
Las Vegas, nv
Apr 84
m IfrB—< d faWMWXI
ftlBBS Mi Maiaite S«ffc*
-------
PE8U- 1784 ^5
EPA-600/4-84-028
April 1984
MULTIELEMENTAL ANALYTICAL TECHNIQUES FOR HAZARDOUS WASTE ANALYSIS:
The State-of-U'e-Art
by
J. A. Oppenhelmer, A. D. Eaton, and L. Y. C. Leong
James M. Montgomery, Consulting Engineers, Inc.
555 E. Walnut Street
Pasadena, California 91101
and
0 % Thomas A. Hlnners
Quality Assurance Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
=y
^ Contract No. 68-03-3076
vsn
: -W
Project Officer
Thomas A. Hlnners
Quality Assurance Division
Environmental Monitoring Systems Laboratory
Ljs Vegas, Nevada 89114
ENVIRONMENTAL MONITORING SYSTEMS LABORATOkY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114
wftoctrto «t
NATIONAL TECHNICAL
INFORMATION SERVICE
IU DIPM1MNI 01 COMHCKCE
¦miwiuo. »* chi
-------
TECHNICAL REPORT DATA
(fltatt nod Instructions on the reverse be fort completing)
1. REPORT SO. i.
ppA-finn/4-R4-n?R
3. RECIPIENT'S ACCESSION NO.
Pffi* 178425
4. TITLE ANO SUBTITLE
MULTIELEMENTAL ANALYTICAL TECHNIQUES FOR HAZARDOUS
WASTE ANALYSIS: The State-of-the-Art
5. REPORT DATE
April 1984
B. PERI ORMINO ORGANIZATION CODE
7.aotmor(s) j. a. Oppenheimer, A. D. Eaton, and
L. Y. C. Leong, Janes M. Montgomery, Consulting
Enalneers. Inc.. and Thomas A. Hinners. EMSL-LV*
B. PERFORMING ORGANIZATION REPORT NO.
B. PERFORMING ORGANIZATION NAME AND ADDRESS
James M. Montgomery, Consulting Engineers, Inc.
555 E. Walnut Street
Pasadena, California 91101
10. PROORAM ELEMENT NO.
ABSD1A
11. CONTRACT/GRANT NC.
Contract No. 68-03-3076
12. SPONSORING AGENCY NAME ANO ADORESS
U.S. Environmental Protection Agency—Las Vegas, NY
Office of Research and Development
Environmental Monitoring Systems Laboratory
Las Yeses, Nevada 89114
13. TYPE OF REPORT AND PERIOD COVERED
Project, 6/81-6/82
14. SPONSORING AGENCY COOE
EPA/600/07
is. supplementary notes *Qual1ty Assurance Division
Environmental Monitoring Systems Laboratory
Las Vegas, Nevada 89114
1ft. ABSTRACT
Based on a comprehensive review of the literature, the mult1elemental techniques
of Inductively coupled plasma optical emission spectroscopy (ICP), x-ray fluorescence
(XRF) and Instrumental neutron activation analysis (INAA) have been compared for the
determination of antimony, arsenic, barium, barylllura, cadmium, chromium, copper,
lead, mercury, nickel, selenium, silver, thallium and zinc 1n hazardous waste matrices.
These particular elements were chosen because they are on the list for classifying a
waste as hazardous or on EPA's Priority Pollutant list. Each technique 1s discussed
with respect to theory, anticipated Interferences, correction techniques, precision,
accuracy, detection Holts and cost. This literature rt
-------
NQTICt
This report has been reviewed 1n accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication. Mention of trade names or commercial products
doc'-i not,, constitute endorsement or recommendation for use.
11
-------
ABSTRACT
Based on a comprehensive review of the literature, the multle'emental
techniques of Inductively coupled plasma optical emission spectroscopy flCf),
x-r*y fluorescence iXRF) and Instrumental neutron activation analysis (INAA)
have been compared for the analysis of antimony, arsenic, barlur, beryllium,
cadmium, chromium, copper, lead, mercury, nickel, selenium, "liver, thallium
and zinc 1n hazardous waste matrices. These particular elements were chosen
because they are on the 11st for classifying a waste as hazardous or on EPA's
Pr1or1cy Pollutant list. Each technique 1s discussed with respect to theory,
anticipated Interferences and correction techniques, observed precision and
accuracy for slqple and complex matrices, detection limits and cost.
The literature review demonstrates there has not been sufficient analyti-
cal work on complex matrices to fully cospare these three methods for many of
the priority pollutant elements. For those elements with a sufficient database
to compare precision and accuracy by the three methods (arsenic, barium,
chromium, lead, nickel and zinc) 1CP has the advantages of lower detection
limits and precision, whereas XRF may be very useful as a preliminary screening
technique due to Its ability to provide rapid serai-quantitative data even at
trace levels. XRF and ICP have significant cost advantages over INAA, requir-
ing much less capital expenditure and lower labor costs.
111
-------
CONTENTS
Abstract. . . . 111
Figures . vHi
Tables. 1x
Abbreviations and Symbols x1
Introduction 1
Conclusions . 2
Recommendations 3
Mult(elemental Methods . . 4
Background . . 4
Available Methods. . 4
Isotope Dilution Mass Spectrometry (1DHS) 4
Spark Source Mass Spectrometry (SSMS) 5
Mufti elemental Atomic Absorption (HAAS) 5
Yoltawnetry (VM) 5
Inductively Coupled Plasma Optical Emission Spectroscopy (ICP) ... 6
Instrumental Neutron Activation Analysis (INAA). ... 6
X-Ray Fluorescence (XRF) 7
Methods of Choice 7
Comparison of Methods 7
Literature Search . 7
Data Management 3
Data Selection 9
Project Approach 9
Detection Limits . 11
Inductively Crupled Plasma Optical Emission Spectroscopy (ICPh .... 1*
Theory 14
Instrumentation 15
Pneunatlc Nebulizers 15
Ultrr.sonlc Nebulizers 16
Spectrometers 17
Poljchromators fSimultaneous Systems) 17
Monjchromators (Sequential Systems) ..... 18
Conparlson of Simultaneous and Sequential Systems 13
Detection Limits 19
Accuracy and Precision 20
Reported Accuracy and Precision . 21
Sources of Error 23
Spectral Interferences 23
Background Shifts 25
Chemical Interferentes. . 25
Physical Interferences 25
Instrumental Drift 25
Preceding page Wank
-------
CONTENTS (Continued)
Correction Methods 26
Alternate Lines 26
Correction Factors • 26
Background Correction . . 26
Sample Preparation ... 26
Fusion 27
Acid Digestion 28
Costs 28
Capital Investment 28
Expendables 29
Operator Skill 29
Sample Throughput 29
Commercial Availability 29
X-Ray Fluorescence Spectroscopy tXRF) . . 31
Theory 31
Instrumentation 33
Wavelength Dispersive 33
Energy Dispersive 33
Detection Limits 34
Theory 34
Observed Detection Limits 34
Accuracy and Precision 35
Sources of Error 35
Sample Preparation 37
Interferences 37
Correction Methods 40
Matrix Matching 40
Correction Factors 40
Modified Correction Factors 41
Transmission and Fluorescence 41
Scattered Radiation ..... 42
True Background 42
Other Methods 43
Summary of Correction. Methods 43
.sample Preparation 44
Liquids 44
Sol Ids 46
Costs 47
Capital Investment 47
Expendables 47
Simple Throughput 48
Operator Skill 48
Commercial Services 48
v1
-------
CONTENTS (Continued)
Instrumental Neutron Activation Analysis (INAA) 49
Theory 49
Instrumentation SO
Detection Limits 51
Accuracy ana Precision 54
Sources of Error 54
Spectral Interferences 54
Absorption Errors 56
Dead-Time Loss 56
Counting Statistics Errors 56
Correction Methods 58
Sanple Preparation 58
Costs 58
Capital Investment 58
Expendables 59
Sanple Throughput 59
Operator Skill 59
Commercial Services 59
Coiqparlson CO
Introduction 60
Detection L1rc1ts 60
Accuracy and Precision . 61
Interferences 62
Flexibility 65
Sample Preparation 66
Cost 66
Throughput 66
Availability 66
Conclusions 66
References 68
Appendices
A. Instrument User Questionnaire 86
B. List of Individuals Receiving the Instrument User
Questionnaire 89
v11
-------
FIGURES
Mumber Page
1 Database sorted by technique, detection limit (DL), precision
(P) and accuracy (A) 10
2 Concentric pneumatic nebulizer ..... 15
3 Flow shear pneumatic nebulizer 17
4 Ultrasonic nebulizer 18
5 Example of a polychromavor-based I CP system 19
6 Example of a monochromator-based ICP system. 20
7 Direct spectral overlap of As and Cd emission lines 24
8 Partial spectral overlap of Cd and A1 emission lines 24
9 Intensity versus wavelength of x-rays transmitted through an
absorber 32
10 Spectral and background radiation Interferences 1n XRF 38
vl 11
-------
TABLES
Number Page
1 Comparison Summary of Instrumental Methods 8
2 Number of Records 1n Database Sorted by Technique and
Element 11
3 Measurement Limits for a Normal Distribution 13
4 Median of Reported ICP Detection Limits for Simple Aqueous
Matrices and Complex Matrices 21
5 Mean Precision and Accuracy by ICP ZZ
6 Spectral Interference Factors Reported for ICP 27
7 Measurement Limits for a Polsson Distribution 34
8 Median of Reported XRF Detection Limits ( 35
9 H?an Precision and Accuracy by XRF 36
10 Spectral Interferences 1n XRF 39
11 Limiting Assumptions for Using Backscatter 1n XRF Matrix
Corrections 44
12 Matrix Absorption Correction Techniques for XRF 45
13 Nuclides Available 1n INAA for the Selected Elements 51
14 Medium of Reported INAA Detection Limits 53
15 Mean Precis'on and Accuracy by INAA 55
16 Principal Interferents 1n INAA for the Selected Elements .... 57
17 Ranking Conqparlson of ICP, XRF and INAA 60
18 Comparison of Median Detection Limits for ICP, XRF and INAA. . . 61
19 Percentage of Cases with Acceptable Precision 62
1 x
-------
TABLES (Continued)
Page
20 Percentage of Cases with Acceptable Accuracy 63
21 Ranking of Techniques by Element 64
22 Ranking of Error Sources for the Techniques 64
23 Analytical Definitions of Random and Systematic Errors 65
x
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
IDMS -- Isotope dilution mass spectrometry
SSMS — spark source mass spectrometry
MAAS -- multielement atomic absorption spectroscopy
VM — vol tamrietry
ASV — anodic stripping voltammetry
ICP -- Inductively coupled plasma optical emission spectroscopy
XRF — x-ray fluorescence
TNAA -- instrumental neutron activation analysis
AAS — atomic absorption spectroscopy
CL -- critical level
OL — detection limit
QL -- quantification limit
RSD -- relative standard deviation
IUPAC — International Union of Pure and Applied Chemistry
R -- reaction cross section 1n INAA
NA -- number of target nuclei 1n INAA
SYMBOLS
a — standard deviation
s -- signal
b -- blank
o -- type 1 hypothesis testing error (false positive) 1n statistics
0 type 2 hypothesis testing error (false negative) 1n statistics
Kfli -- abscissa of standardized normal distribution corresponding to
probabllHty levsl l-o 1n detection limit discussions
Ko — emission 1n XRF for electron transition from L atomic shell
to K atomic shell
KB -- emission In XRF for electron transition from M atomic shr>ll
to K atomic shell
La -- emission 1n XRF for electron transition from M atomic shell
to L3 atomic subshells
Kg -- abscissa of standardized normal distribution corresponding to
probability level 1-B In detection limit discussions
um — mass absorption coefficient
^ -- neutron flux density
T(l/2) — elemental halt-life
Xg -- decay constant of radionuclide
\1
-------
SECTION 1
INTRODUCTION
Information on the pollutant content of wastes Is required to properly
assess the hazard and the need for waste management or for remedial action.
Multlelemental analytical techniques are attractive as the Beans to obtain
Inorganic content Information on wastes rapidly and/or economically. Altnough
nultlelemental techniques are widely ussd, complex wastes and contaminated soil
present the potential for interferences not normally encountered. The elements
ant1mor\y (Sb), arsenic (As), barium (Ba), beryllium (Be), cadmium (Cd), copper
(Cu), chromium (Cr), lead (Pb), mercury (Hg), nickel (N1), selenium (Se),
silver (Ag), thallium (Tl), and zinc (Zn) are identified In federal publica-
tions as pollutants of special concern.
The present work is a literature evaluation of currently available multl-
elemental techniques 1n order to compare the advantages and limitations for the
analysis of the specified elements 1n hazardous wastes or 1n contaminated
soils. The currently available imi1t1ele
-------
SECTION 2
CONCLUSIONS
The results of a preliminary evaluation of all currently available multl-
elemental methods Indicated that Inductively coupled plasma optical emission
spectroscopy (ICP), x-ray fluorescence (XRF), and Instrumental neutron activa-
tion analysis (INAA) are most amenable to development as routine methodologies
for analysis of complex matrices for the 14 elements.
A rigorous comparison of these three techniques was made on the basis of
detection Units, precision, accuracy, freedom from Interferences, flexibility,
sample preparation, cost, thoughput, and availability. All three techniques
have the capacity for detecting numerous elements In wastes at concentrations
greater than 100 ppm. XRF could be useful as a p'-ellminary screening technique
since 1t produces semi-quantitative results 1n 5 minutes or less. Overall, ICP
appears to be the best method for determining these particular elements 1n
hazardous waste matrices, but our recotaaendatlon 1s not conclusive because 1t
1s based only on published data.
2
-------
SECT'.OH 3
RECOMMENDATION
A sore relevant comparison of the three techniques .sfrsttM be based ca
Measurements for the 14 elewtnts 1n the saoe tM2ardous waste samples. Use of
published data does not provide an adequate database far coaparlson of trie
aetftod^ because very MUle wort has been published cm nut tl elemental analysis
of hazardous waste «atricef for this set of eleoents. The database Is, there-
fore, Insufficient for tuny of these eleaents; and deta for the three setftod-
ologles are not directly coaparable because thejp are derived frt* different
aatrlces and based on d1f!>rent detection Halt definitions. The database may
alio be Inherently biased because published data ire usually superior to data
derived frew routine applications.
3
-------
SECTION 4
MULT I ELEMENTAL METHODS
BACKGROUND
Congress enacted the Resource Conservation and Recovery Act in 1976 to
protect the public ana the environment from improper disposal of hazardous
wastes. Under this legislation the EPA is required to develop a nationwide
program to regulate hazaroous waste practices. EPA is required under Section
3001 of this Act to identify all solid wastes which must be managed as hazard-
ous wastes and to identify the specific characteristics of each hazardous
waste. Included in this monitoring program are measurements for 14 elements
consisting of the 13 inorganic priority pollutants (Ag, As, Be, Cd, Cr, Cu, Hg,
Ni( Pb, Sb, Se, T1, Zn) and Ba. Barium has been classified as a hazardous
constituent by the EPA.
Presently, several multielemental techniques are candidates for simulta-
neously determining these pollutants in hazardous waste samples. Development
of routine trailtielemental techniques for hazardous wastes could provide rapid
analyses and substantial economic savings. It has not been clearly demonstrated
whether the time savings (from multielementa" capability) outweigh the capital
investments and analytical 1 Imitations that may arise from attempting to simul-
taneously determine 14 elements in complex matrices. For example, many multi-
elemental techniques still cannot be used to analyze solid samples directly
without preparation; and a single preparation technique may not be adequate fcr
all 14 elements of interest. This report assesses the potertial of currently
available multielemental techniques for determining the specified pollutant
elements in hazardous waste samples. Each technique is evaluated in terms of
accuracy, precision, detection limits, severity of matrix interferences and
economic factors.
AVAILABLE METHODS
The currently available multielemental methods are isotope dilution mass
spectrometry, spark source mass spectrometry, multiel. nental atomic absorption
spectroscopy, po'.arography, inductively coupled plasma optical emission spec-
troscopy, instrumental neutron activation analysis, and x-ray fluorescence
spectroscopy.
ISOTOPE DILUTION MASS SPECTROMETRY (IDMS)
In isotope dilution mass spectrometry the concentration of an element is
determined by observing the change which occurs in its natural isotopic composi-
tion upon tie addition of isotopic tracer (of that same elenent) which has an
4
-------
artificially altered composition. IDMS exhibits good precision and Interfer-
ences occur only when Isotopes of different elements have the sane mass nun&er.
Separation of Interfering elements can usually be achieved within the 1on
source If these elements have different volatilities. The major drawbacks of
tills technique are the very expensive capital Investment, the need to Introduce
san^les Into an evacuated chamber and the long analysis time (compared to other
available methods). It cannot determine the 17 monolsotoplc elements, which
Include the pollutant As; and the working concentration range 1s narrow because
1t Is a ratio method. When the concentration of the element of Interest 1n the
sample 1s low, the composition of the mixture containing both simple and tracer
will approach that of pure tracer and become almost indistinguishable from It.
When elemental concentration In the sample 1s high, the mixture composition
approaches that of pure sample and once again creates the problem of 1nd1st1n-
gu1shab1l1ty. The usabU concentration range for each particular element 1s
dependent on the 1sotop1c ratios naturally occur!ng for that element. When the
natural ratio 1s close to unity, as 1n the case of Ag, the working range Is
extremely short.
SPARK SOURCE MASS SPECTROMETRY (SSMS)
In spark source mass spectrometry an electrode, which 1s partially or
completely composed of sample, 1s subjected to a pulsed radio-frequency voltage.
The high voltage spark which Is produced causes localized heating of the elec-
trode. The sample then vaporizes as atoms or simple 1ons which are accelerated
through an 1on gun and mass analyzed. Elements are Identified by the mass-to-
charge ratio of their 1ons and concentration Is determined from the Ion current.
The main advantage of spark source mass spectrometry 1s the extreme specificity
and low detection limits' of the technique for almost all elements. Its major
limitations are essentially the same as those for Isotope dilution mass spec-
trometry; namely, the high cost of Instrumentation, requirement for a skilled
technician, and long analysis time. In addition, electrode preparation would
probably requlrr digestion of a waste and adsorption onto an electrode.
Another potential source of trouble Is the use of an extremely small sample
size, which may lead to errors In accuracy If the analyzed aliquot 1s not repre-
sentative of the bulk sample.
MULTIELEMENTAL ATOMIC ABSORPTION SPECTROSCOPY (MAAS)
Multlelemental atomic absorption spectroscopy 1s based on conventional
atomic absorption spectroscopy which has been modified to achieve simultaneous
element measurements through use of a continuum source and an echelle mono-
chromator modified for wavelength modulated detection The main advantage of
this technique 1s Its flexibility, maintaining the advantages of conventional
AAS with respect to specificity and freedom from Interferences. Its major
limitations are a restricted linear working range, poor detection lfmlts com-
pared to AAS, and lack of commercial availability (which would lead to large
capital expenditures 1n setting ur a laboratory for MAAS).
YOLTAMMETRY (YM)
Yoltammetry 1s an electro-analytical multlelemental technique which con-
sists of a cell containing a dropplng-mercury Indicator electrode, a reference
5
-------
electrode, a counter electrode and the solution of Interest. When a gradually
changing vo'itage 1s applied to tins cell, ions are reduced at the Indicator
electrode 1n the fonn of a mercury amalgam. This reduction process results 1r<
the production of a cell current which 1s directly proportional to the solution
concentration of the element ov Interest when the current 1s diffusion con-
trolled. Element Identification 1s achieved by observing the potential at
which the curreot Is half the magnitude of the diffusion controlled current
(half-wave potential). Very low detection limits can be achieved utilizing a
specialized form of voltammetry known as anodic stripping voltansnetry (ASV),
but 1t 1s only applicable to a few netals and sample pretreatment 1s required
when determining total metal as opposed to labile complexes. The major draw-
back to voltanaetry Is that element Identification 1s strongly dependent on a
knowledge of the matrix composition because the element half-wave potential
shifts 1n response to other species present 1n the sample matrix. Therefore,
highly skilled technicians are required to adequately perform ASV. Further-
more, multielemental determinations can only be made 1f the half-wave poten-
tials of the metals of Interest are sufficiently far apart for the current to
be measured accurately. These types of severe interference problems Indicate
that voltammetry does not possess adequate selectivity to be useful as a multi-
elemental technique for the determination of metals 1n complex matrices.
INDUCTIVELY COUPLED PLASMA OPTICAL EMISSION SPECTROSCOPY (ICP)
Inductively coupled plasma 1s a form of optical emission spectroscopy
which utilizes argon plasma as the excitation source. This transforms the
sample Into electronically excited Ions and atoms which emit photons as they
deactivate to their ground state. The quanta are sorted by wavelength 1n a
spectrophotometer and the resulting errfsslon 1s detected photoelectrically.
Signal response Is proportional to concentration level and each element emits
11ght of a characteristic wavelength. This technique possesses several advan-
tages. Linearity can be expected over 4-6 orders of magnitude, detection
limits are low, costs are moderate, analysis time 1s fairly rapid, and the
requirement that s»ples be Introduced as liquids ensures obtaining results
which will truly represent the bulk sample. The primary disadvantages are
matrix Interferences and the fact that solid smples cannot be analyzed
directly. Dissolution of the sanple introduces the potential for contamination
or loss of analyte, and decreases detection Units.
INSTRUMENTAL NEUTRON ACTIVATION ANALYSIS (INAA)
Instrumental neutron activation analysis 1s a radiochemical method 1n
which the sample is exposed to a neutrcn flux and the induced radioactivity 1s
determined using gsnma ray spectrometry. The advantages of this technique are
Its nondestructive nature, minimum sample manipulation requirements and low
detection limits. The main limitations are matrix Interferences, required
access to a neutron source with associated high capital costs, the fact that
not all elements are amenable to this type of analysis, and the possibility
that the small solid specimen wnlch 1s analyzed may not be homogeneous and
representative of the bulk sample.
6
-------
X-RAY FLUORESCENCE (XRF)
X-ray fluorescence 1s a form of x-ray spectroscopy In which a sample 1s
bombarded with x-rays and Inner orbital electrons are ejected from atoms within
the sample. The resulting excited atoms dissipate energy by filling these
vacancies with electrons from higher energy levels and emitting characteristic
x-ray photons. This secondary emission cf x-rays 1s referred to as fluores-
cence and because only certain emissions occur, the x-ray spectral line Is
Indicative of the element present. Element concentration 1s deduced from the
x-ray count rate of a characteristic line. The advantages of this method are
basically the same as those of neutron activation analysis. XRF 1s nondestruc-
tive, allows for direct analysis of solid samples, and exhibits low detection
limits. Disadvantages of this methodology are the severity of matrix effects,
difficulty In analyzing liquid samples directly, and the fact that elements
with atomic nrnber less than 10 cannot be determined.
METHODS OF CHOICE
Table 1 sumurizes the advantages and disadvantages of each of the poten-
tial analytical methods. Isotope dilution mass spectrometry and spark source
mass spectrometry possess unacceptable economic factors of large capital cost,
expensive labor costs, and lengthy analysis time. Multielemental atomic absorp
tion spectroscopy 1s not currently available commercially and has not been
fully Investigated. Polarography lacks sufficient selectivity to be developed
as a method for routine multielemental analyses of complicated matrices.
ICP, INAA and XRF appear to be the three most viable currently available
multielemental methods which could he developed as routine methodologies for
determining the specified elements 1n hazardous wastes.
COMPARISON OF METHODS
As a result of the conclusions reached in the preliminary evaluation, the
next sections of this report will provide an extensive coiiparlson of ICP, INAA,
and XRF as methodologies for determining the specified elements 1n hazardous
wastes. This comparison 1s based on analytical data obtained from a thorough
literature search and from replies to a questionnaire sent out to users of the
various techniques. The questionnaire 1s shown 1n Appendix A. The con|>ar1son
provides an assessment of each technique with respect to analytical capabili-
ties and limitations, detection limits, precision a,id accuracy, complexity of
sample preparation, availability of commercial services and cost.
Literature Search
A thorough literature search was accomplished with the assistance of the
Lockheed computer system DIALOG which enables the user to search selected
abstract flies. The files searched Included Chemical Abstracts, NTIS, Cotiq>en-
dex (collection of engineering Journals), Agricola (collection of agricultural
journals), Foundation Grants Index, Dissertation Abstracts, Envirollne (collec-
tion of environmentally oriented journals), Scisearch, Pollution Abstracts,
SSIE Current Research, Conference Papers Index. More than 50 different
7
-------
TABLE 1. COMPARISON SUMMARr OF INSTRUMENTAL METHODS
{~'advantage, -^disadvantage)
sasefldasMSsaaasaaaaaasssjsa ajassaaaaE'-LussoaaaasE? saaassssssays&aaaa asaa staaava
MAAS AAS ICP XRF INAA IDHS SSMS V
+
+
+
+
+
+
*
+
+
+
Minimal Operator Skill +
Required
Cost (Capital, Operation and
Maintenance)
Detection Limits +
Freedom from lnterferencas +
Precision +
Availability
Working Range
Large Number of Elements +
Analysis Hoe +
aax>*Q aaiaaasoaaaisasa aaa«3saaawa»:
HAAS ¦> Multielement*1 Atomic Absorption Spectroscopy
AAS * Atomic Absorption Spectroscopy
ICP » Inductively Coupled Plasma Optical Emission Spectroscopy
XRF » X-Ray Fluorescence
IHAA * Instrumental Neutron Activation Ana^sls
IDBS ¦ I«otope Dilution Mass Spectrometry
SSHS a Spark Source Mass Spectrometry
V ® Voltaametry
+
+
+
+
+
+
aaxcaa ¦ aia^ataaava
+
+
+
+
^>^.3 avaasaa s s a x saa a sis s«is:s a
journals and abstracts were searched. The key papers used In this review are
cited in the references which follow the 1CP, XRF and INAA sections.
Data Management
Introduction—
Any publication i^1ch provided Irfomatlon on detection limits, precision,
or accuracy for any of the selected elements was Included in the data manage-
ment system, which was accomplished with ths aid of the computer system Data-
trieve on the VAX 11/780 computer. Each elemental analysis of a sample
consisted of a sepa'ate record in the system so that any group of elements
could be combiinsa. Thus, a reported sample analysis could produce up to 14
records for a given methodology if an article included information on precision,
accuracy, or detection limits for all 14 elements. Typically, a given journal
article would produce about 10-?0 records depending on the unber of samples
analyred and the number of elements determined. The advantage of Datatrleve 1s
tnat large amounts of data can be grouped in any fashion desired to obtain
selected Information; I.e., it Is possible to readily examine the literature
and determine »rttic"i techniques provide the raoct informat.on on any given ele-
ment and which techniques have the best precision, accuracy or detection
limits.
8
-------
Data Selection
Availability of Data--
The available precision, accuracy, and detection limit data determined for
any glve.i element by ICP, INAA or XRF are summarized 1n Figure 1. Data for
conventional (single-element) AAS was not collected and summarized because the
precision, accuracy and detection limits obtained by AAS are already well
established. Inherent 1n this summary 1s the assur,i^tion that the literature
search was reasonably complete. Each record represents one piece of data on a
given element. It becomes clear from examining this table that the number of
journal articles where ICP has been used for determining the specified elements
far exceeds the database available for neutron activation or x-ray fluores-
cence. This reflects, in part, the fact that INAA and XRF have traditionally
been used to determine a different set of elements than ICP.
Table 2 provides a breakdown of the data on an element and technique
basis. This provides a rapid way of comparing the available data for certain
elements. For Instance, a mlnltoal amount of data was found for x-ray fluores-
cence analysis of Ag, Cd, Sb, Hg, Se and T1. XRF has traditionally been used
for analyses of major elements and this explains the paucity of data for these
aetals which typically occur at fairly low concentrations 1n environmental
samples. For neutron activation there 1s a minimal amount of data available on
Be, Pb or T1 because these elements do not form adequate gamma-emitting radio-
nuclides. For ICP, little data are available for Hg or T1 because these
elements can be determined to lower concentrations using cold-vapor AAS and
furnace AAS (respectively), and few standard reference materials have these
elements present at concentrations measurable by ICP. This database provides
the basis on which subsequent comparisons of the three methodologies are made.
PROJECT APPROACH
The database can be sorted for each element to determine:
• concentration ranges which can be analyzed with adequate accuracy and
precision
• levels of accuracy and precision which can be expected in complex and
variable hazardous waste matrices
9 obtainable detection limits 1n liquid and solid matrices
t types of Interferences
9 correction methods used
t sample preparation techniques
Following this assessment of analytical capabilities, data gathered from
direct user contact and response to our questionnaire was used to estimate:
• capital expenditures
• labor costs
• supply costs
t throughput rate
• required operator skill
9
-------
2.049
Figure 1. Datatnse sorted by technique, detection limit (Dl),
precision (?) and accuracy (A).
10
-------
TABLE 2. NOMPSS OF RECORDS IN DATABASE SORTED BY TECHNIQUE AND ELEMENT
tasaflauaaaasvsaasaasasasiasiaaaaasssaaaassassaaaasssassaaassassssassssasssssaia
Method
Element TCP TTOR W Total
Ag
34
10
1
45
As
50
18
11
79
Ba
87
53
82
222
Be
5*
1
0
56
Cd
110
15
7
132
Cr
137
54
96
287
Cu
198
12
66
276
Hg
9
18
4
31
N1
149
34
116
299
Pb
166
6
56
228
Sb
27
31
1
59
Se
27
18
3
48
n
7
0
0
7
Zn
164
35
78
277
aaaaaaMasaaasaBaBsassaasaasaaaaaasasaiaaaaasBaasssaaaaaasnsaasffSsaassssaaaBatis
Finally, the relative merits of each riethod were balanced against the
availability of commercial services for the techniques and the availability or
standard analytical protocols.
Thorough consideration of the analytical and economic variables for each
Instrumental technique should make it possible to assess the technlque(s) best
suited for determining the 14 specified elements 1n hazardous wastes. Each
technique was evaluated for Its potential as a screening procedure for unknown
hazardous wastes as well as for Its potential to produce highly accurate and
precise data.
DETECTION LIMITS
In using our database to compare ICP, XRF and INAA, It became apparent
that the definition of detection limit (DL) used by various workers, especially
in ICP, varied tremendously. A survey of the literature reveals that a number
of definitions are being used to define the detection limit. Some work 1s
based on two or three times the standard deviation of the background noise
while other papers take two or three times the standard deviation of either a
low concentration standard, the ordinate intercept of the calibration regres-
sion or a matrix Identical to the sample (but not containing the element of
Interest). This 1s significant 1n analysis of hazardous wastes where matrices
can be tremendously complex and present numerous interferences. In assessing
the data we have not differentiated these oeflnltlons. However, we believe 1t
1s Instructive to present a statistical basis for the proper definition of
detection limits so future work can be readily compared.
11
-------
An excellent statistical derivation of detection limit definitions for
normal and Polsson distributions 1s given by Currle" and surenarlzed below.
The critical questions asked about detection limits are the following:
1. Given an observed signal, has a true (or non-random) signal been
detected?
2. What 1s the smallest analyte level which will dependably yield a
detectable observed signal?
Using hypothesis testing, we want to minimize the error (o) of saying an
observed signal Indicates a true signal when a true signal 1s not present.
Choosing the confidence one wishes to have 1n excluding false positives fixes
the value of a. A critical level (CL) can then be determined from the chosen a
value and the blank standard deviation (<75), equation 1.
CL = Ko • ah (1)
This critical level Indicates a signal that 1s probably not a random event.
The inverse error (3) of saying a signal 1s not present when 1t actually
1s must aUo be minimized. Here, one wants an estimation of the true signal
which will yield an observed signal which 1s sufficiently qreater than the
background noise to allow detection. The detection limit \DL) 1s therefore
dependent on CL, 8, and the detection-level standard deviation (
-------
If the standard deviation Is approximately Independent of signal level,
equation 4 reduces to equation 5 for the case of paired samples and blanks.
In this situation the number of sample observations and blank observations are
equivalent. Equation 6 applies if the blank is well-know) due to a long
fclst'jry of observations.
_ 2 -2 <51
? s = 2<7£
a 2 , -2 16)
*s a b
Table 3 sunaarlzes the expressions which are valid for any normal distribution.
lCP-der1ved data follows a normal distribution and, consequently, the
detection limit [recommended by Currie25) 1s (at the 95 percent confidence level)
3-3 times the standard deviation of a blank (at the 95 percent confidence level}.
The 1UPAC recoimwnds3b the use of 3^ for detection limit determinations at the
90 percent confidence level.
XRF and INAA involve counting radiation processes which follow a Polsson
rather than a norwal distribution. The detection limit definition for a Polsson
distribution 1s given 1n the XRF section. In the Individual sections on ICP,
XRF and INAA, the various detection ?imit definitions utilized in reporting
data, and their Inherent assumptions, are discussed.
TABLE 3. MEASUREMENT LIMITS FOR A NORMAL DISTRIBUTION*
ssa3sa9ass83saa3»33aRssas3aa33ssssB38saa3s3ass3388339S8asss3assss3aassa9ssa333a
Critical Detection Quantification
Level (CD Limit (DL) Limit (QL)
Paired sample and blar.k 2.334.65*b W.l»b
"Well-known" blank 1.64*b 3.29
-------
SECTION 5
INDUCTIVELY COUPLED PLASMA OPTICAL EMISSION SPECTROSCOPY (ICP)
THEORY
Optical emission spectroscopy 1s a technique which Involves sample absorp-
tion of energy (from a flame, spark, arc or plasma) to produce a population of
excited atoms and 1ons which return to the grourd state by emitting radiation.
The wavelengths of the emitted photons are characteristic of the elements
present because of the quantum mechanical rules wnlch govern these electron
transitions.
The excitation source plays an Important role 1n the stability, detection
limits and Interferences of optical emission spectroscopy. Only a fraction c "
the atoms within a sample are excited. This fraction 1s dependent on the
temperature of the excitation source. A high temperature will Increase the
population which 1s excited and improve the detection limits of the technique.
The source should also have good stability to prevent fluctuations in the
population of excited atoms and, therefore, ensure stable spectral-11ne Inten-
sities. Materials 1n the excitation source can contribute to Interference
effects 1n optical emission spectroscopy.
Historical excitation sources are either gas combustion flames or arc
sources on graphite crucibles. Combustion flames contribute considerable
chemical Interferences due to reactions which occur with the elements contained
1n the flames. Arc sources on graphite crucibles have poor detection limits
for a number of elements due to Incomplete excitation and to consumption of the
graphite by the arc source (which contributes cyanogen Interference bands).
Inductively coupled plasma optical emission spectroscopy (ICP) utilizes
Inductive coupling of time-varying magnetic fields with an argon plasma instead
of conventional gas flames. A plasma possesses the properties cf a gas out '
contains sufficient positive ions and electrons to conduct an electric current
and interact with macnrtlc fields. The aruon is initially made conductive by
the discharge of a Tesla coll which introduces seed electrons within the elec-
tric field. The Induced axial magnetic fields cause the charged particles of
the plasma to flow 1n closed annular paths. The electrons and 1ons meet
resistance to their flow resulting 1n Joule heating and additional ionization.
A tangentlally Introduced argon flow cools the walls of tne quartz tubes con-
taining the high-temperature plasma and helps sustain the plasma. A third
argon flow transports the sample, as an aerosol, through the center of the
annular shaped plasma where the optimum temperature range occurs.
14
-------
Plasmas have several advantages over conventional gas flames 1n optical
emission spectroscopy. Lower detection limits are obtained with plasmas than
with other emission sources due to the longer residence time of the sample
within the excitation zone, which has a higher temperature and much cleaner
background spectrum. Atomlzatlon occurs 1n the high-temperature core where
continuum emission 1s quite high, but the atoms are released downstream tc a
lower-temperature environment where a relatively clean background spectrum
occurs. There are fewer chemical Interferences due to the relatively Inert
atmosphere of plasmas conpared to conventional flames and a much larger linear
working range of four to six orders of magnitude due to the high temperature of
the plasma center which prevents self-absorption effects.
INSTRUMENTATION
The Instrumental coi^jonents of ICP consist of a nebulizer (which aspirates
the sample and generates an aerosol Into the plasma) and a spectrometer (which
resolves the emitted radiation). There are two different classes of nebulizers
commercially available for aerosol generation and two types of spectrometers on
the market.
Pneumatic Nebulizers
Pneumatic nebulizers consist of glass, or metal, capillaries where the
rate of aerosol production 1s dependent upon the carrier-gas flow.
1. The concentric pneumatic nebulizer depicted 1n Figure 2 Introduces
the sample through an Inner concentric capillary while the nebulizer
argon flows through an outer tube. The constriction In the solution
capillary 1s frequently clogged by particles that may be 1n the
sample.
A
Figure 2. Concentric pneumatic nebulizer.
15
-------
Two accessories (described below) are available for concentric nebu-
lizers which reduce the problem of nebulizer clogging wien samples
wltn high-salt concentrations are analyzed. These dev'ces Involve
a) Bubbling the carrier argon through water prior to Its passage
through the nebulizer.
b) Rinsing the nebulizer tip by automatically Injecting water
through the nebulizer at the end of each analytical cycle.
These two devices permit aspiration of solutions with dissolved or-
ganic content as high as 10 percent and dissolved Inorganic content
as high as 20-30 percent. Aerosol carr1er-gas flows are typically
1n the range of 1 11ter/m1nute. Since the aerosol carrier gas Is
also used as the nebulizer gas when utilizing pneumatic nebullzatlon,
this required low-flow rate Imposes restrictions on the dimensions of
nebulizer orifices.
2. Figure 3 provides an example of a flow shear pneumatic nebulizer which
operates by directing a solution by gravity flow across a surface con-
taining the gas exit port. There 1s no solution capillary to clog
because only the nebulizing gas goes through the exit port to create
the aerosol.
3. A crossflow pneumatic nebulizer (consisting of 2 capillary tubes that
terminate at 90" to each other) avoids problems of clogging by use of
a sample capillary which is not constricted at the tip. This type of
nebulizer can handle aqueous salt solutions with concentrations
greater than 25 percent. Concentric nebulizers are generally "flore
liable to clog at high-salt concentrations than crossflow nebulizers.
Ultrasonic Nebulizers
A second category of nebulizers are the ultrasonics depicted 1n Figure A.
In ultrasonic nebullzatlon, an aerosol 1s produced by exposing a liquid to
ultrasounds. Typically, this Involves pumping a liquid onto a transfer plate.
Unlike the pneumatic nebul'zers, ultrasonic nebulizers allow the aerosol pro-
duction and the carrier gas flow rate to vary Inlependently thereby optimizing
performance. The sample Injection rate with ultrasonic nebulizers 1s 5-10
times higher than the rate for a crossflow pneumatic nebulizer. Ultrasonic
nebulizers produce denser aerosols and inore uniform particle size which lowers
detection limits. Although the ultrasonic nebulizer Improves the detection
limits, 1t does have some disadvantages. Any matrix effects which are present
will Increase 1n magnitude as the sample Injection rate Increases and memory
effects are more severe. Desolvatlon of the aerosol 1s necessary prior to Its
entry Into the plasma and thfs may cause Interferences. These drawbacks plus
the higher cost Involved 1n utilizing ultrasonic nebullzatlon tend to outweigh
Its advantages over pneumatic nebulizers.
16
-------
Samp* Introdwctton Tube
Figure 3. Flow shear pnetroatic nebulizer.
Spectrometers
Two different kinds of spectrometers are available for application to
multielemental determinations. The isolation and characterization of the
spectral lines may be achieved either with a simultaneous polychromator or a
sequential monochromator.
Polychromators (Simultaneous Systems)
A polychromator consists of an entrance slit, a diffraction grating and
multiple exit slits which are located on the focal curve. Each exit slit is
aligned to a specific elemental emission 11ns and has its own photomultipller
detector and electronic channel. Conmerclally available polychromatic spectrom-
eters are usually equipped with channels for 40-50 emission lines (Figure 5).
Gn Cjui tori
t-wgacio* Rod Mount
17
-------
Aerosol
aivc connecw
Figure 4. Ultrasonic nebulizer.79
Monochromators (Sequential Systems)
A scanning monochrcmator consists of only one photomultlplier detector,
and determines the elements of Interest by moving sequentially from one wave-
length to another {Figure 5}.
Comparison of Simultaneous and Sequential Systems
The polychromators lack the flexibility of the sequential monochromators.
The analysis tine for the polychromators, however, 1s shorter than that of the
ncrtochroMtors. Analysis time for a simultaneous system 1s <2 minutes for 30
elements compared to 1-2 elements/Hnute for a sequential system. Recent
developments In scanning mechanisms for sequential systems have increased the
throughput tw '• elements/minute. When determining a broad range of element
concentrations <"t widely varying sample matrices, the fixed exit slits that are
used for Isolating spectral lines 1n a simultaneous system may become overly
13
-------
Figure 5. Example of a polychromator-based ICP system.
restrictive. The standard spectral Ifne may not be usable due to severe Inter-
ferences from other elements which are present. In this type of situation, an
alternate line may have to be selected at a sacrifice 1n detection limit. A
programmable scanning rnonochrotoator facilitates thts type of procedure while
the option for choosing the appropriate line with the polychromator will not be
available unless a scanning mechanism Is p^rt of the system. Polychromators
are thus the ideal Instrument only when analyzing routine samples of uniform
composition for a selected group of elements.
DETECTION LIMITS
for the purpose of this report, we have ignored any differences between
the definitions of detection limit in susmarizing the dsta. By comparing the
median detection limits observed for each element, we can avoid some of the
ambiguity inherent in different detection limit definitions. Hediar detection
limits for simple aqueous matrices (such as deionixed water or acidified
water) are shown in Table 4. The volatile elements As, Hg, Pb, Sb, Se, and T1
all have median ICP detection limits which are above 20 ug/1. As, Hg, Pb, Sb,
and Se are the elements traditionally determined by flameless AA procedures.
The median detection limits obtained for more complex matrices (such as acid
digests of sediments, steal, or coal fly ash) are listed also In Table 4 and
are not significantly different from those observed in simple matrices. This
is in part an artifact of the variable definitions of detection limits stated
earlier. As discussed in the Hulti el emeriti! Methods Section, the detection
limit should be based on a partlcula* matrix. However, this sumnary provides
19
-------
Figure 6. Example of a monochromator-based I CP system.
a good indication of the magnitude of the detection limit for the different
matrices for each element. The relative independence of ICR detection limits
from matrix effects indicates the utility of this technique for analysis of
wastes.
ACCURACY AND PRECISION
Having demonstrated the detection limits typically achieved with ICP in a
variety of matrices, it was important to assess the accuracy and p^cision of
the technique at different concentration levels under various conditions.
Accuracy and precision are affected by a number of factors:
1. Sample preparation of non-aqueous material may affect the recovery of
a given element as a result of Incomplete digestion, volatilization,
or contamination.
20
-------
TABLE
.jIAN OF REPORTED ICP DETECTION LIMITS FOR SIWLE AQUEOUS
MATRICES AMD COMPLEX MATRICES*
Element
Aqueous Detection
Limit (ug/1)
Complex Matrix Detection
Limit (ug/1)
Ag
As
Ba
Be
Cd
Cr
Cu
K9
Ml
Pb
Sb
Se
T1
Zn
6 (12)
30 (15)
1 (15)
1 (14)
2 (23)
4 (19)
3 (23)
20 (8)
7 (19)
30 (24)
50 (10)
50 (12)
60 (5)
3 (22)
4 (4)
39 (3)
1 (6)
0.2 (2)
2 (10)
4 (12)
2 (13)
30 (1)
8 (15)
20 (14)
30 (2)
18 (1)
23 (1)
2 (16)
•Parentheses contain the nunber of reported detection Units.
2. Once a sample 1s In a homogeneous solution, errors can arise fro*
spectral interferences which may be caused by direct spectral overlap
between the element cf Interest and an Interfering element or a
background shift. Examples of these interferences are cited below
and, In a more detailed discussion of Interferences, we will enumerate
the reliability of available correction methods.
3. Accuracy can be readily affected by errors In preparation of stan-
dards. This 1s an error which 1s coMon to any analytical method and
therefore can he considered Independent ef technique.
First, the reported accuracy and precision comprising the database 1s
discussed and then the 1nterferences and correction methods for ICP follow.
Reported Accuracy and Precision
Table 5 shows the mean precision and accuracy by ICP reported for each
element as a function of concentration level. Accuracy 1s assessed by examin-
ing percent recoveries for analytes added to samples or by agreement with
certified values for reference materials, such as Standard Reference faterlals
(SRMs), from the National Bureau of Standards. For 12 of the 14 elements there
tre sufficient data available to make some assessments for determinations by
ICP. For Hg and T1 there are no available data on precision and accuracy.
This Is principally because Hg and T1 are usually determined by cold-v^bcr AAS
and furnace AAs, respectively. In addition, there are no SRMs with t./tlfied
values for Hg of T1 at levels measurable by ICP. For the remaining elements,
the precision of ICP as reported In the literature 1s fairly Independent of
21
-------
concentration level and typically being better than a 15 percent relative
standard deviation. The only exception is As which exhibits poor precision at
high concentrations {10-1000 mg/1). This 1s probably due to the fact that As
1s volatile and easily lost during sample preparation. The accuracy of 1CP 1s
excellent with most data Indicating recoveries between 80 and 120 percent
(Table 5).
TABLE 5. MEAN PRECISION AMD ACCURACY Bf ICP
taaasosaaaaaaaifassjsssasaa^asasstassaasassaflaasssaa^ssss^aaisaaaisssassassaBfs
Concentration Precision Accuracy
Element Range (nj/gl {percent RSO] [percent recovery)
*g
Below I
5
lf8
1-10
3
99
10-100
11
150
AS
1-10
5
91
10- 100
19
98
100-1000
-
97
04
10-100
5
91
100-1000
8
303
Above 1000
4
99
Bo
1-1G
e
106
10-100
16
69
100-1000
14
112
Cd
Below 1
12
134
s-ia
2
112
10-100
5
100
Cr
1-10
7
84
10-100
5
105
100-1000
5
94
Above 1000
1
B3
Cu
Below 1
50
100
1-10
14
98
10-100'j
7
98
Hg
HA*
HA
HA
Ml
Below 1
22
92
1-10
17
92
10-100
B
125
100-1000
4
97
(continued)
22
-------
TABLE 5. (Continued)
BaaaasiiBaaaaucasaiaaaaaa93aai333saattr.a98aaaass3a38as3GBsaaaaa9i538BeaB«S3S3aass
Concentration Precision Accuracy
E.ement Range Ug/g) (percent RSD) (percent recovery)
Pb
Below 1
14
96
1-10
17
104
10-100
8
120
L00-1CT0
5
142
Above 1000
2
98
Sb
10-100
9
MA*
Se
1-10
NA
81
10-100
HA
90
T1
NA
NA
NA
In
Below 1
11
88
1-10
1
107
10-100
8
106
100-1000
6
100
Above 1000
4
9S
Baaaafla«3saia3aa*a3sa3a»aaa8aaisa»asaa«3a3a8sasa3:3aaa933is3saasa3aaaaaiassa«s
*NA = not available
SOURCES OF ERROR
Accuracy is potentialiy affected by a number of error sources. Analytical
errors 1n ICP are caused by direct spectral and background interferences due to
matrix elements, sample interaction with the plasma, variable physical charac-
teristics of different matrices and Instrianent drift.
Spectral Interferences
Spectral interferences occur when the energy of the emitted radiation of
the element of Interest cannot be resolved from the radiation emitted by matrix
elements of the sample. A direct spectral Une overlap is observed and a false
high rending Is obtained. An example of this 1s the 228.B0-n»n spectral Une
for Cd which directly overlaps the 228.81-mn line of As (Figure 7).
Overlap may also occur even when the energy of the Interfering spectral
peak Is more than a ma from the spectral peak for the element of interest (If
this Interfering element exhibits a very broad peak). Overlap will then occur
between the element of interest and the wing from the interfering element.
An example of this type of Interference Is the 214.44-nm line of Cd with the
215.54-nm line of AI (Figure 6).
23
-------
0.1 pg/mLCd
226.802 nm
10 pg/mL As
^228.812 nm
I
1
1
2285 nm
229.1 nm
Figure 7. Direct spectral overlap of As and Cd emission lines.28
214.0 215,0 nm
figure 8. Partial spectral overlap of Cd and A1 emission lines.28
24
-------
Background Shifts
The spectral Hne for an element nay also be affected by background
emission. Stray light from st>-ong emission lines and brocd contlnuion radia-
tion can produce high spectral readings. Such background shifts can occur
for numerous elements when wajor constituents are at high concentrations.
Chemical Interferences
Chemical Interferences are due to the Influence of certain reactions
occurring In the plasma. These cause disturbances, either by affecting the
Ionization of the element of Interest or by forming refractory compounds with
the test element which affect the production of atoms and ions for emission.
Due to the clean environment of plasmas, only the Ionization type of Interfer-
ence 1s a problem. When the element of Interest 1s easily Ionized and the
sample matrix contains high conceittratlons of easily Ionized concomitants, a
Mgier reading for an atom line will occur due to a shift 1n the ionization
equilibrium of the analyte.
The extent of this effect 1s dependent upon the Ionization potentials of
the elements of interest, the relative concentrations of these elements and the
Instrumental operating conditions, the observation height 1n *he plasma, and the
power dissipated 1n the plasma. The selected instrumental settings are usually
a compromise between mitigating chemical Interferences and roax.mizlng detection
limits. The extremely high temperature of the plasma does not support exten-
sive Ionization Interferences because the plasma is electron-rich and provides
a well-buffered environment. Optimization of the radio-frequency power, gas
flow, and observation height are usually sufficient techlques to minimize this
type of interference.
Physical Interferences
Physical Interference occurs when there 1s a difference in the aspiration
rate of samples and standards Into the plasma. Unless the rates of nebullzatlon
for sample and standard are identical, the accuracy of the result will be ques-
tionable. Variability in the acid or salt concentration from sample to standard
will affect the accuracy because the viscosity, surface tension, and density
will differ from sample to standard. Detection limits are also degraded when
analyzing samples with high acid or salt concentrations due to deterioration of
the nebulizer performance. To ensure prevention of this type of physical error,
the acid and salt content of the calibration stindards should be carefully
matched to the samples. If matching is not possible, an internal standard
should be used and the appropriate correction factors applied
Instrumental Drift
Instrumental drift involves any changes In the performance of the instrument
as a function of time. This factor was Included 1n our analysis based on
responses to our questionnaire. User responses to the question on instrianental
drift were quite variable. The replies rargod from statements that drift was
as high as 12 to 15 percent during the course of an 3-hour day to statements
that drift was not a problem. Recallbratlon 1s performed from two to four
25
-------
times dally by those users reporting Instrumental drift. Drift was attributed
to nebulizer Instability and temperature effects.
CORRECTION METHODS
Alternate Lines
The optimal correction for the problem of spectral-11ne overlap 1s to
switch to an Interference-free analysis line for the element of Interest. In
the example of As Interference on Cd, alternate lines Include the 214.44- and
226.50-nm Cd lines. Capability to shift to a different analysis line requires a
sequential spectrometer or a scanning mechanism for a simultaneous Instrument.
Alternate emission lines may not provide sufficient detection limits for some
applications. Use of Interference filters on simultaneous Instruments 1s not
usually sufficient to remove severe spectral Interferences.
Correction Factors
If switching to a different analysis line 1s not possible, computer
corrections can be made. The concentration of the Interfering element can be
discerned at an analysis line other than the one which overlaps. Analysf of a
pure standard of the Interfering element will then Indicate the ratio of ak
heights at these two different analysis lines. Once this information 1s
obtained an empirical spectral Interference factor can be calculated which will
correct the measured peak height for the Interfering element. Tables of these
enpirlcal correction factors have been compiled. Currently there 1s a dispute
1n the literature as to whether these factors can be linearly interpolated for
Interferent concentrations. Church^3 claims the correction factors are linear
up to 1000 mg/1 of Interferent while Fassel^ maintains that correction curves
should be established. Table 6 displays the major Interferences' and empirical
Interference factors reported by Church23 for the elements of Interest.
Although Interference factors depend upon the particular Instrument used, this
table 1s Indicative of the magnitude of effects.
Background Correction
Computers are also utilized to perform background corrections for analyte
peaks. The technique consists of measuring the background Intensity at a
certain distance on one or both side(s) of the analyte peak and subtracting the
background Intensity from the analyte peak Intensity. Tht: wavelengths used for
background measurements must be selected with care to achieve proper correction.
if the Interference problem 1s so severe that none of these procedures
provide adequate analysis, then the only remaining alternative is some type of
chemical separation such as solvent extraction or hydride formation.
SAMPLE PREPARATION
Liquid samples (oraanlc or aqueous) require no preparation orlor to
analysis by ICP unless the sample contains suspended particulate matter or a
high concentration of dissolved solids. Particles must be filtered out or
26
-------
TABLE 6. SPECTRAL INTERFERENCE FACTORS REPORTED* FOR ICP
aaiaaaaBaassasassaaaasssssasasaisaasasssssssasassssssssssssssssssssssssssssssar.
Element Interference Factor"1"
(wavelength) Major Interferent (ppb/ppm)
Ag (328.1 nm)
Zr
6 +
As (193.8 nm)
Y
7
Ba (455.4 nm)
Zr
0.3
Be (313 mn)
V
0.6
C<< (266.5 ntn)
N1
0.3
Cr (283.6 nm)
Fe
1
Cu (324.8 nm)
Be
1.3
Hg (253.6 ran)
Co
10
N1 (231.6 nm)
Co
0.1
Pb (220.4 nm)
Co
1
Se (196.1 nm)
V
1
Zn (202.6 nm)
Cr
0.8
asaaaaaaaaaaaaaaaaaaassaaaaaaaasaaaaaaasaaaaaaaasaaaaasaassaaaaassasaaaaassaaaa
* Data from Church23
+ Factor Indicates that a 1-ppm Zr solution will contribute emission at 32E.1 nm
equivalent to 6 ppb of Ag.
digested If they are to be Included 1n the analysis. Difficulty 1n analyzing
high-salt solutions 1s due to clogging of the nebulizer. Procedures for
alleviating this problem are discussed under "Instrumentation". For conven-
tional ICP solid samples require a dissolution step prior to analysis. Sample
preparation and homogeneity are the ".argest sources of error 1n ICP when an
adequate nimber of recallbratlons are performed during the day to correct for
Instrumental drift, the acid and salt matrices of samples and standards
are matched, and the operator 1s experienced enough to recognize and correct
for the spectral and background Interferences. The major digestion procedures
are fusion and add digestion.
Fusion
Fusions are typically performed with bicarbonate, LIBO2 or NaOH followed
by acid dissolution. For routine analysis of trace elements, fusion 1s Imprac-
tical due to manipulative difficulties, the Introduction of alkali metal salts
27
-------
(which may clog the nebulizer or cause undesirable matrix effects, simple
dilution, analyte losses, potential for high blanks, and added expense uje to
the required high purity materials.
Acid Digestion
Acid digestion utilizes heat and either a single acid, or some appropriate
combination of acids, to dissolve the sample. Hydrofluoric (Hr), hydrochloric
(HC-1), nitric (HNO3) and perchloric {HCIO4,) are the acids typically employed.
The particular combination of acids selected Is a function of the type of
sample matrix being digested. For soil contaminated with hazardous waste
samples, nitric acid may be sufficient, but the combination HF/HCIO4/HNO3 will
ensure complete digestion. Thompson5* and others have found a simple HNO3
digestion to be greater than 90 percent effective in solubillzing trace metals
such as Cu, Pb, Zn, Cd, Ni, rig and Ag from clay minerals. In additiot to the
concern that dissolution may be Incomplete, loss of the volatile elemt'its As,
Hg, and Se must be prevented. This is usually achieved by carrying out the
digestion procedure 1n sealed Teflon boo>bs.
The advantages of sample dissolution are Insurance of homogeneity ano the
ability to select extraction schemes which may mimic real-life processes such
as selective leaching from a dump site (e.g., dilute-acid leaching}. Total
digestion may not reflect environmental availability of toxic elements. The
disadvantages are the additional expense, the potential for contamination,
dilution, patential loss of volatile analytes, the potential for Incomplete
dissolution and the explosive hazard of HCIO4. Furthermore, when using acid
digestion, care must be exercised to ensure that all samples and standard:
contain the same percentage of acid since acid content w1l' af:'cct the rate of
sample nebulizatlon and, therefore, the accuracy of the results.
COSTS
ICP costs are a function of the number of samples analyzed. The capital
expense can be amortized to a srall increment ot unit costs If "here Is a large
number of samples. Expendables are minimal and labor 1s the largest single
cost on a per sampla basis.
Capital Investment
J50K (3as1c ICP) Add $10K for sampler and printer
$100K (ICP/AAS full system - moncchroirator or 20 channel simultaneous)
|175K. {Top-of-the-11ne simultaneous with scanning monochronwtor and 40
channels).
There are several major U.S. manufacturers:
Jarrell-Ash - only polychromator unit
28
-------
Instrumentation - only monochrooator unit
Laboratory
Applied Research - Doth monochrooator and polychromator units
Laboratories
Perkln-Elraer - only monochromator unit
Balrd - only polychromator unit
Instruments, S.A. - both monochromator and polychromator units.
At the present time, Intense competition and developments 1n ICP have
reduced capital costs markedly. It 1s now possible to obtain a high quality
sequential or simultaneous system for under $60K.
Expendables
The major expendable supply 1n ICP 1s argon (which Is consumed at a rate
of 1-2 I1ters/m1nute). Users Indicate typical costs for the gas plus other
supplies (standards, etc.) to be about $l-2/samp1e.
Operator Skill
Operator skill requlrements for ICP are significant. Although the tech-
nique Is presented as being rapid and straightforward, the operator must be
aware of the various types of corrections required and the Importance of each
one. The ability to properly establish, maintain, and operate a computer
system which handles spectral and background Interferences comes only with
experience. Thus, labor costs are the most significant factor 1n estimating
total ICP costs.
Sample Throughput
The Instrumental analysis rate Is 30-200 samples/day. The rate-Hmltlng
step for the analysis of solid samples 1s the required preliminary digestion.
Although the digestion can be set-up for batches of 50-100 samples, the overall
digestion time for these samples 1s several hours.
Coirmerdal Availability
Many laboratories offer ICP analysis at rates ranging from |20-100/sample,
depending on the number of elements being determined and the complexity of the
sample preparation. Host commercial laboratories do not typically define
detection limits on a matrix-dependent basis and thus may provide overly opti-
mistic detection limits. Responses to the user questionnaire indicate that
many laboratories may not *>e familiar with all the potential problems Involved
In ICP analysis of wastes. Although methods have been published for ICP
analysis by EPA (eclted by Theodore 0. Martin and John F. Kopp, Environmental
Monitoring and Support Laboratory Office of Research and Development, U.S. EPA,
Cincinnati, OH 45268, EPA 200.7, Interim Method - Federal Reglstei , December
3, 1979) and ASTM (1981 Annual Book of ASTM Standards, Part 31, Hater.
29
-------
American Society for Testing and Materials, 1916 Race Street, Philadelphia,
PA), these methods are geared to waters with minimal Interferences, ant comner-
dal laboratories without well-trained personnel cannot automatically apply
these procedures to difficult matrices to obtain reliable data. A round-robin
study on 1CP analyst of wastes would be useful to establish the reliability of
commercial services.
30
-------
SECTION 6
X-RAY FLUORESCENCE SPECTROSCOPY (XRF)
THEORY
A sample target bombarded with sufficiently energetic x-rays *111 eject
Inner-shell electrons from atoms within the sample, ''or this process to occur,
the x-rays must possess a minimum energy corresponding to the excitation
potential to remove the appropriate photo?lectron. The resulting excited atoms
dissipate this energy by losing a second electron or by emitting radiation.
Electron loss 1s referred to as the Auger effect while photo emission 1s known
as x-ray fluorescence.
When fluorescence occurs, a particular element will emit a photon with a
characteristic wavelength because quantun selection rules limit the possible
electron transitions. A count of the emUted x-ray Intensity can be used to
calculate the quantity of the element present 1n the sample. The wavelength of
a photon emitted by a particular element Is Inversely proportional to the square
of the atomic number of the element. Light elements emit x-rays that havo
wavelengths which are too long to be dispersed by conmerciall.y available
analyzing crystals. This coupled with the fact that radlationless transitions
Increase with decreasing atomic number makes x-ray fluorescence analysis of
light elements with atomic number less than 10 Impractical.
The minimum excitation radiation energy to remove an Inner shell electron
Is dependent on the energy level shell of the electron as well as the atomic
number of the element. K shell electrons require the largest amount of energy
for removal because these shells are located closest to the nucleus of the
atom. A plot of the transmitted Intensity of the bombarding x-rays as a func-
tion of their wavelengths (Figure 9) reveals a series of sharp spikes. Each
spike is referred to as an absorption edge and will occur where the incoming
radiation does not have sufficient energy to remove electrons from a particular
electron-energy level.
Calculation of an element's concentration from Its x-ray count rate Is not
a straightforward procedure 1n a mixed element matrix because the other elements
present absorb some of the bombarding x-rays. heasurswnt of the decrease
1n the intensity of the bombarding x-rays as they attempt to pass through the
target sample gives an Indication of the amount of x-ray absorption occurring
within the sample (equatlor 7).
(7)
-------
I
K
3
vJN
V
wavelength
Figure 9. Intensity versus wavelength of x-rays
transmitted through an absorber.
where
I » the Intensity of the x-ray beam of an Initial Intensity I0
x = the thickness of the absorber
u » the linear absorption coefficient
The linear absorption coefficient, when normalized for the density of the
absorber, yields m, (which 1s known as the mass absorption coefficient). Each
element has its own characteristic mass absorption coefficient and the absorp-
tion coefficient increases with Increasing wavelength of the bombarding photons.
A plot of the mass absorption coefficient as a function of the wavelength of
the bombarding photons yields a curve which 1s the inverse of the transmitted
Intensity curve given In Figure 9. The mass absorption coefficient for a
particular element can be determined from equation 8.
Z * the atomic number
A » the atomic weight
n * an exponent ranging In value from 2.5-3
C ¦ a constant which changes at each absorption edge
X ¦ wavelength
The total mass absorption coefficient for a sample 1s an additive function
of the mass absorption coefficient of each individual element present 1n the
sonple multiplied by that element's weight fraction (equation 9).
(8)
where
Mm ¦ wAWA + WBHB
32
(9)
-------
where
H ¦ element weight fraction
w ¦ mass absorption coefficient for each element
The relationship between elemental concentration and measured character-
istic x-ray count rate, given In equation 10, 1s complex because it Is dependent
on a knowledge of the natrlx absorption correction. In order to obtain accurate
values with XRF, the absorption correction values for a sample must be accu-
rately known, empirically derived or minimized.
PPK(1> - Ci/Cs x Ab(corr-i) * x 1/kf (1C)
wher«
PPM(1) • concentration of element i In parts-per-mlllion
Ci » x-r«y court rate from element 1
C« » x-ray count rate from element standard
Ab(corr-1) • %/^"e"Uni " a^sorpt1on correction for matrix
absorption effects
Ms ¦ mass of standard
Ki ¦ relative excitation - detection efficiency for
x-ray lire from element 1 compared with standard
element x-ray line.
INSTRUMENTATION
XRF requires a spectrometer to sort the wavelengths of the photons which
are emitted from the bombarded sample... Two spectrometer types are currently
avail alle.
Wavelength Dispersive
One type is the wavelength dispersive x-ray spectrometer 1n which the
fluorescent x-rays are diffracted by a suitable crystal according to Bragg's
Law. Wavelength dispersive spectrometers were the original type developed,
but they have been superceded by energy dispersive systems due to cost con-
siderations.
Energy Dispersive
Non-dispersive x-ray spectrometers operate by using semiconductor detec-
tors to sort fluorescent x-rays according to their differences In energy. The
nondlsperslve type offers the advantages of easier operation, more rapid anal-
ysis, less expense and avoidance of the use of crystal optics which can de-
crease x-ray Intensities through radiation loss. The disadvantage of this
Instruaent the poorer resolution obtained at energies less than 10 keV.
Progress In electronic design over the last few years has significantly im-
proved this energy resolution. The advent of guard-ring detectors has dras-
tically reduced the x-ray spectrum background. Currently, two types of semi-
conductor detectors are 1n use. Ge(H) detectors offer high resolution but
33
-------
must be operated at liquid nitrogen temperatures, whereas Hgl detectors show
great potential for operating at room temperatue but are not yet developed with
adequate resolution for trace level work.
DETECTION LIMITS
Theory
A survey of the current literature reveals that the computation of detec-
tion limits for XRF 1s usually based on the definition given by Currie."
Currie states that the processes Involving counting follow a Polsson rather
than a normal distribution (unless the number of counts Is large enough to
approximate a normal distribution). In a Polsson distribution, the standard
deviation cannot be considered Independent of signal level (unlike ICP where
the collected data follows a normal distribution). A summary of the quali-
tative and quantitative measurement limits for the situations of paired (sample
and blank) data or a well-known blank are given in Table 7.
TABLE 7. MEASUREMENT LIMITS* FOR A POISSON DISTRIBUTION*5
aaaaaaaisaaavaBSsaaaaaaaasssBSSkaassassaanaBiaaiisaanaaaaaaflaasiaasaasaaasisaBsa
Critical Limit Detection Limit Quantification Limit
(CL) (DL) (QL)
Paired sample 2.33 Wb 2.71 * 4.65 ub 50 i1+ fl+ Wb 1 1/2 I
and blank ' I 12J5J J
"Wei 1-Known" 1.64 Wh 2.71 + 3.29 wh 50 h+ |"l+ Wb 1 1/2 I
blank ° ° ' L 25 J >
laaaaaaaaaaaaaaaaBaaaaaaaaBaaaiBBaaaaaaaaaaaaaBaaaaaaaasaaaaaaaBassasaaaaaasaaa
•When a ¦ 8 ° 0.05 (for 51 risks), Kg » 10, and ub ¦ ne*" value for blank
Use of double-guard ring detector and pulsed x-ray tube can improve detection
limits by approximately a factor of 4.
Observed Detection Limits
Table 8 shows the median detection limits computed for each element. Con-
ventional XRF cannot determine Be, and no data were reported for Ag, Sb or T1.
These limits were obtained for various matrices and the detection limit defini-
tion Is usually based on Curr1e.25 Uniformity of the detection limit definition
1s not assured because some workers assume the definition based on a Polsson
distribution, others assume a normal-distribution approximation and use two or
three times the standard deviation of the background, while others use the
definition based upon a Polsson-distrlbutlon bjt assume the absence of system-
atic and non-Poissonian random errors. For the specified elements, the detec-
tion limits ?re about 1-2 ug/g with the exceptions of Cd (5 ug/g) and Ba (10
ug/g). The limits obv»rved for Cd and Se are questlonnable median values
because they are based on only one piece of data each. The poor detection
34
-------
TABLE 8. MEDIAN OF REPORTED XRF DETECTION LIMITS*
siaaiaaaiiauasisaaassaaasaasstssssaaisasssaasssasaassssraaaassssaasaaaaaaaaaaa
Detection Liir.1t Detection Limit
Element (vg/g) Element (ug/g)
Ag
NA
Hg
2 (3)
As
1 (2)
Ni
1 (4)
Ba
10 (3)
Pb
2 (6)
Be
NA
Sb
NA
Cd
5 (1)
Se
13 (1)
Cr
1 (4)
Ti
NA
Cu
1 (6)
Zn
3 (5)
saaiaaaiaaaasaaaaaaaaaaaaaaaaaaiaaaaaaaaaaaaaasaassaisssasssi
~Parantheses contain the number of reported detection limits
NA - Not Available
Unit for Ba may be due to thu fact that Its fluorescent x-rays have the same
energy as a large fraction of the background radiation.
ACCURACY AND PRECISION
Given stable operating conditions, constant detector geometry and adequate
background calibration, XRF Instrumental precision becomes a function of count-
ing statistics only. This source of error can be Isolated because It 1s not
significantly influenced by the manner 1n which the spectral data 1s accumu-
lated and processed. A counting error rarely exceeds d few percent ana typi-
cally does so only 1f the detection limit 1s approached. Analytical precision
for XRF therefore appears to be limited by operator error. Under constant
conditions the major source of Imprecision will be due to the error 1n dupli-
cating sample preparation rather than Instrumental Instability. In XRF, errors
1n sample preparation exceed 1% and arise from the Inability to make pellets,
fusions sr thin films of uniform mass thickness.
The mean precision and accuracy for each element as a function of concen-
tration level is shown in TaDle 9. Two trends are apparent in Table 9. Pre-
cision improves with increasing corcantratlon, and accuracy 1s independent o'
concentration except at the extremely low concentrations where It is quite poor
for sane elements. Both of these observations are the expected ones. The pre-
cision does not seen to reach an acceptable value (<20%) until the concentration
1s greater than 10 ug/g and 1n some cases (Ba, Cr, Ni) not until it Is greater
than 100 ug/g* Barium and chromium, along with Pb, a-e also the elements that
exhibit poor accuracy at lower concentrations. The precision and accuracy
results for Sb and Se will not be interpreted becau'e they are based on only
one piece of data each.
SOURCES OF ERROR
One of the major problems present with every multlelemental technique 1s
the influence of matrix effects. These are often severe 1n XRF because a sig-
nificant portion of the secondary ar.d primary x-rays are absorbed by the major
35
-------
TABLE 9. MEAN PRECISION AND ACCURACY BY XRF
aaBasasaBasuasaaBSftBsssaaaasyasaasBssissaaflussssBsaissaaasdssassasaBSsaaaaisatas
Concentration Range Precision Accuracy
Element (u9/g) (percent RSD) (percent recovery)
Ag
NA
NA
NA
As
1-10
37
90
10-100
8
98
Ba
10-100
34
122
100-1000
6
105
Above 1000
4
100
Be
NA
NA
NA
Cd
1-10
NA
91
10-100
NA
80
Cr
1-10
51
142
10-100
25
166
100-1000
10
96
Above 1000
3
98
Cu
1-10
89
98
10-1C0
18
103
100-1000
5
10b
rig
NA
NA
NA
N1
1-10
72
105
10-100
47
103
100-1000
8
97
Above 1000
3
100
Pb
1-19
NA
229
10-100
12
103
Sb
1-1G
75
58
Se
10-100
9
109
71
NA
NA
NA
Zr,
1-10
NA
91
10-100
6
99
100-1000
12
98
Above 1000
3
137
73333333333388891333333^^33883383333333333 33388533333233:338338833833833333X389
-------
constituents comprising the sample matrix. The key to obtaining reliable
x-ray fluorescence data 1s proper sample preparation. The sample must be
prepared 1n a manner which will minimize matrix absorption problems and also
ensure homogeneity (which Is Important because Incident x-rays do not penetrate
far Into the sample). Conversion of x-ray count rate to element concentration
will not be accurate unless the sample mass absorption effects can be compen-
sated for or eliminated.
Sample Preparation
Solid samples may be prepared In accordance with any of three different
procedures.
• Fusion
• Pelletizatlon
• Filtration
Fusion of a sample with lithium tetraborate or borax will reduce matrix
effects and ensure homogeneity, but loss of volatile elements and sample dilu-
tion with subsequent reduction 1n detection limits) are undesirable for trace
element analysis.
If fusion 1s not utilized, the sample must be finely ground 1n order to
overcome particle size effects. Uniformity can be assessed by grinding until
further grinding gives no charge In the fluorescent x-ray intensity. This
powder can then be mixed with compressed borax or salts of light elements
which do not produce an Interfering spectrum and pressed into pellets or else
slurried and filtered onto a M1111pore-type filter. Preparation of pellets
Involves a minimum amount of time and ensures homogeneity but does little to
alleviate matrix effects. The filtration method is more time-consuming and
produces less stable samples (flaking can be a problem unless an acrylIc-spray
coating 1s applied) but tMn, uniform filtered samples do have less severe
matrix 1nterferences than pelletlzed samples.
Interferences
Interferences which can occur 1n addition to matrix absorption are direct
spectral Interferences, background radiation Interferences, and self-absorp-
tion of fluorescent radiation (which Is referred to an enhancement). Typical
x-ray fluorescence spectra are shown 1n Figure 10 and both spectral Interfer-
ences and background radiation interferences are apparent. The Ba peak is
located on the Incoherent scattered background radiation from the excitation
source, and several peaks occur on shoulders of adjacent peaks with higher
fluorescent Intensities.
Spectral I nterference—
Direct spectral Interference occurs when a spectral line from one element
directly overlaps the spectrum line of another element because the emitted
photons are of nearly equal energies. An example of this Is the Kcq line
emitted by T1 which Interferes with the first order Laj line of Ba. The LBi
line of Ba does not have a T1 Interference, but the Lai line from Ce will
Interfete. The region below 4 keV contains the K lines of elements up to Ca
37
-------
Figure 10. Spectral and background radiation interferences in XRF.^6
and the L and M lines of the heavier elements. Several interferences occur in
this region, notably Zn with Na and Pb with S. However, since the Zn L x-ray
and Pb M x-rays are not normally used fo.- analysis this 1s a problem which is
more of theoretical interest. Heavier elements can usually be identified by
using the higher energy lines in the same spectrum, [n the region of 4-8 keV
interferences from the I. lines of Ba and the K lines of Ti affect the other
elements which emit photons in this region, such as Cr. The V Kb line and the
x-ray tailing background from the large K-line emission of Fe (usually present
1n geological type samples) also Interfere with Cr. In the region above 8
keV the only significant interference is the Ka line of As with the La line
of Pb.
Corrections for these types of interferences can be attempted by adjust-
ing the pulse-height selector to minimize the interference effect or using
secondary lines to correct for the interferent. If neither of these two pos-
sibilities is viable, correction for the interferent may be initiated by com-
puter applied empirical correction factors which calculate the relationship
between the interfering intensity and another line of the interfering element.
Table 10 summarizes the potential Interferences for the 13 elements of interest
measurable by conventional XRF.
Background Scatter--
Backgrourid radiation is principally the result of scattered excitation
radiation. The scattered radiation seriously affects the detection limits
of the technique by decreasing the ratio c
-------
TABLE 13. SPECTRAL INTERFERENCES IN XRF
Msiasas93saaass3saaaass2 aaaaaaaa ^a^sas sa«^as aasss^aaaas^ss s
Analytical line Interference
Ag Ka
As Ka
Pb La
to La
Ti Ka
3d IB
Yi K3
3a Ka
Cd Ka
Cr Ka
v Kb,
Cr KB
Mn Ka,
Cu Ka
N1 Kfl
Hg La
Ge Ka
Kg Lfl
Br Kg
Hi Ka
Fe Kg
Pb la
AS Xa
pb la
Sb Ka
Se Ka
T1 La
Pb la
f! Lb
Br Ka
Zn Ka
Cu Kb
V Ka, Ce la
Fe K, Tailing Background
Tailing Background Ba l
Zn Ke
aaa»a9aaasaasaRSsasa«s33Sc;.3»Ba533as333ss«3ii»3ssas8aB89sese»a83SQaa3as53S3a
monochromatic excitation radiation should be used. Conventional x-ray tubes
arid Brenrnstrahlung sources give a continuum of interfering scattered radiation
and, consequently, low x-ray peak intensity relative to background. The angle
between the exciting and detects radiation also influences the intensity of
the scattered radiation. Although the energy loss by Incoherent scattering
increases tilth the angle cf scattering, at an angle of 90® the intensity of the
scattered radiation is mf/Jima! relative to the fluorescent radiation Inten-
sities. An angle of 90® ts therefore most favorable because it maximizes the
peak-to-bockground ratios. The energy chosen for the excitation radiation also
has an affect oi> the amount of scattered radiation thai; occurs. Maximum detec-
tion limits are obtained when the excitation radiation is slightly greater than
the K or l absorption edge energies of t.ne elements tn be determined, but of
sufficient energy that scattered radiation does not- produce significant over-
lapping background.
Background scattar is also a function of the specimen thickness. Because
absorption effects Increase more rapidly for fluorescence than scattered exci-
tation radiation, detection limits decrease as specimen thickness approaches
the critical thickness {the thickness beyo. d which an increase in thickness
does not Increase fluorescence x-ray Intensity), Maximum detection Units are,
therefore, achieved by using thin specimens. The reduced counting rates which
are obtained from thinner specimens can b* compensated for by using a more
intense excitation radiation. In the case of multfelemental analysis, flu-
orescent x-rays will cover a broad range of energies and a trade-off in the
39
-------
ratios of x-ray line Intensities to spectral background for soma of the ele-
ments will occur.
The spectral background for elements which have K x-rays in the range
31-32 keV (e.g., Ba) contain a larger fraction of scattered radiation which
increases the errors in their determination. These errors can be reduced and a
higher x-ray line spectral intensity achieved by operating the x-ray tube at
higher voltages. Thir. will shift the scatter radiation to higher energies,
rhis may not be possible, however, if the excitation system is limited to
certain maximum voltages.
Computer correction may be applied for background radiation interference
effects. Some commercial systems are equipped with software to compute the
background correction factors for each peak so that the background need only
be measured at one position during the analysis of unknowns.
Enhancement--
Enhancement effects are maximized when the major component of a sample
emits a wavelength just shorter than the absorption edge of the element of
interest. Experimental results have indicated that enhancement effects
are negligible for most thin specimens.
CORRECTION METHODS
Matrix absorption effects can be minimized by using thin sample specimens.
Both t.hin pelletized samples ana thin targets collected on filters still have
some finite thickness, however, and yield data which require corrections.
Matrix Matching
The simplistic approach for dealing with matrix absorption eliminates the
need to quantify and correct for this error by ensuring that identical absorp-
tion occurs in the samples and standards. This can be achieved through the
use of standard additions, Internal standards, or matrix matching standards to
samples. Matrix matching 1s the most impractical of these methods since it
requires quantitative knowledge of the sample constituents and hazardous waste
samples are typically complex anJ highly variable. Use of an internal stan-
dard is less time consuming but a qualitative knowledge of the matrix is still
necessary since the element introduced as the standard must be absent from the
sample. Knowledge of sample matrix is not needed for standard additions but
the multiple aeterminations required for each element make this procedure
extremely time-consuming when determining 13 elements per sample.
Correction Factors
Tables of empirical Interaction coefficients (k-j, j) generated from the
analysis of standards containing one najor component and one trace element can
be utilized for matrix absorption corrections. Use of these coefficients in
equation 11 enables calculation of a concentration corrected for absorption
effects.
40
-------
n
CC, * I-f £ » J) Cj
0=1
(U)
where
I-f ¦ intensity of element of interest
CC^ » corrected concentration for element of interest
Cj * concentrations of interfering elements
Adequate corrections can be made provided the exact matrix composjtion of
the sample is known. This technique is not viable for analysis of hazardous
waste samples because the matrices are highly variable and major element
composition is typically ur':nown.
Modified Correction Factors
A modification of the empirical interaction coefficient technique allows
absorption corrections to be made without any knowledge o* the exact major
elemental composition of the sample. Iteration of equation 12 on n+2 stan-
dards (where n equals the number of Interfering elements) allows the determi-
nation of the concentration of element i corrected for the absorption effects
of the other sample elements.
CCi => A + 8 If [exp {-ki, j Cj)] (12)
This method is inconvenient because it still requires qualitative knowledge
of the sample matrix composition, and the preparation and analysis of n+2
standards per sample is not cost-effective.
Transmission and Fluorescence
This correction method does not depend on quantitative or qualitative
knowledge of the sample matrix and is applicable to hazardous waste samples.
Matrix absorption corrections are calculated from transmission and fluores-
cence measurerants or. pellet samples of uniform mass thickness. This is done
by measuring the relative x-ray intensity with and without '.he sample from a
target located at a postion adjacent to the back of the sample. The mass
absorption correction is calculated according to equation 13.
It
where Is, ly, Ij* ?re the intensities of the x-ray plus background from the
specimen alone, the target alone, and the specimen plus the target, respec-
tively. This technique corrects for 1nter-sa;nple matrix variations and yields
41
-------
accurate results 1f the specimen Is less than the critical thickness. The
requirement of two analyses per sample makes this method time-consuming and
less cost-effective; furthermore, the quantity of raw data produced could
exceed the 30-40 kilobyte core memory of most minicomputers.
Scattered Radiation
Use of scattered radiation is another absorption correction technique
which works for matrices of unknown composition. When monochromatic x-rays
bombard the sample they are scattered as well as absorbed. Two different
types of scattering can occur. Rayleigh scattering is scattered radiation
which experiences no change in wavelength (coherent scattering). Compton
scattered radiation is caused by an elastic collision between a photon and an
electron In which the photon loses energy. The amount of energy lost Is depen-
dent upon the angle between the scattpring direction and the incident beam.
This type of incoherent scattering predominates when the energy of the incident
x-ray photons are much greater than the binding energy of the bombarded elec-
trons. Since the binding energy increases with Increasing atomic number,
scattering by heavier elements 1s mainly coherent. Mass absorption coeffi-
cients are also a function of the atomic number so determination of intensity
of Compton or Raylelgh scattering by a sample can be related to the sample's
mass absorption coefficient. There is a linear relationship between Compton
scattering and ir.bss absorption coefficients for materials of known composition.
Absorption corrections can be based on the Compton scattering, the Ray-
leigh scattering, or the ratio of both. The ratio method may be used with a
continuum or discrete line excitation while the other two require a discrete
line source in order to achieve resolution of the coherent and incoherent
components. Use of Raylelgh scattering either alone or in combination with
Compton scattering, determines a less accurate mass absorntion coefficient,
value due to diffraction of Raylelgh scattered radiation.
The Compton scatter intensity may be used to determine mass absorption
coefficients. The procedure requires that the specimens be of Infinite thick-
ness for all radiations of interest and that the bombarding x-ray radiation be
of shorter wavelength than the wavelength corresponding to the absorption edqe
of the highest atomic number major element 1n the matrix. For geological
samples this is usually Fe and the Compton scattering correction technique can
be used down to the Ka wavelength of Ni utilizing an Au, W or Mo x-ray tube as
excitation source. The ratio of the mass absorption coefficient for the sample
and standard measured at any particular wavelength in the region of wavelengths
shorter than the Fe absorption edge Is constant for all other wavelengths in
that region. Outside of this region, the amount of Fe present in the sample
must be known in order to calculate the ratio correctly.
True Background
For a series of different samples, a plot of an element's spectral back-
ground versus the Compton-scattered radiation 1s linear, but does not inter-
cept the origin. This residual background at the intercept is attributed to
instrumental factors (such as electronic noise as well as scattering and
fluorescence from the analyzing crystal). Subtraction of this resfdual yields
42
-------
the fctrue" background which is inversely proportional to the mass absorption
coefficient. Correction for matrix absorption can therefore be obtained by
taking the ratio of the total peak intensity minus the total background to the
"true" background. The correct concentration of the element of interest can
be calculated from an equation.
At 7.1 keV, the total mass absorption coefficient of a soil-derived sam-
ple changes abruptly because of the Fe K absorption edge. The true background
intensity determined for elements with x-rays <7.1 keV is inversely propor-
tional to the apparent mass absorption coefficient rather than the true coeffi-
cient. Thus, without taking this into account, a serious systematic error
would result if one assumed a uniform change in the mass absorption coefficient
with energy. The proper correction requires the determination of the weight
fraction of Fe present in the sample and the difference between the apparent
and the true total mass absorption coefficient. Utilizing this background
technique, only one background reading is required to calculate both background
and matrix variation corrections for spectral peaks of Interest.
Other Methods
The scattered radiation techniques discern the mass absorption due to the
heavy elements (Z > 13). Since environmental samples are typically composed
of a significant amount of light elements (such as C, H and N) they should be
considered when making corrections. Nielson^ developed a method to correct
for absorption due to heavy and light elements. Thin film calibrations are
used to determine the incoherent and coherent scattering caused by heavy ele-
ments. The remaining fractions of the scatter peaks are then used to deter-
mine the quantity of two representative light elements. Utilizing an iterative
process, the heavy and light element concentrations are used in computing self-
absorption, enhancement, and particle size corrections from which new heavy
element concentrations are determined. This technique avoids the burden of
making transmission measurements on each sample or making assumptions about
sample composition or having to dilute the sample in order to accurately know
its composition. Table 11 outlines the assumptions inherent in utilizing the
different scattering techniques to correct for matrix absorption. The only
assumption inherent in Nielson's technique is that diffraction is negligible.
Summary of Correction Methods
Table 12 summarizes all the matrix absorption correction techniques and
their limitations and disadvantages. Although all of the techniques are ca-
pable of generating good data, the first four techniques listed are not appli-
cable to highly variable complex matrices. Use of these techniques on a rou-
tine basis for hazadous waste analysis would be prohibitively time-consuming
and expensive. The scattered radiation techniques are rapid and most cost-
effective; however, those utilizing Rayleigh scattering may be less accurate
due to diffraction effects. The true background residual technique is pref-
erable to the total background technique because this eliminates the assumption
that detector background is negligible. Any of the remaining scattered radi-
ation techniques should be amenable to routine analysis of hazardous wastes.
43
-------
TABLE 11. LIMITING ASSUMPTIONS FOR USING BACKSCATTER IN XRF MATRIX CORRECTIONS
assasaaaaaassaasaassosssaaaaosssasosssvAaraaaasaaasaassasasssasBssaasBszaasaass
Assumption Method
1. Sample is infinitely thick or of thickness equal
to standards
I, C, (I+C)lIght
B, TB
2* The ratio ^sample analyte line
vstandard
equals the ratio wsample/frsample
ystandard/astandard
at the scatter energy
3. Oiffraction is negligible
4. Background at analyte energy is from the sample
and not due to detector background
5. The "true" and "residual" background intensities
can bo accurately resolved
I. C, (I+C)light.B
!S33Saaiaa9S3S3S3:33:3&SS33S33S33
:s333»aa
C. (I+C)l1ght, B
(I+C)light
B
TB
1339333333333353333
u a mass absorption coefficient
0 = mass scattering coefficient
C = ratio of fluorescent intensity to ccherent scattering intensity
1 * ratio of fluorescent intensity to incoherent scattering
intensity
B = ratio of fluorescent intensity to total spectral peak background
TB = ratio of fluorescent intensity to true background (total back-
ground - residual instrumental background)
(I>C)11ght = ratio of fluorescent Intensity to Incoherent and coherent
scattering including light elements
SAMPLE PREPARATION
Liquids
Sample preparation techniuqes are of prime consideration in obtaining
accuracy, precision and adequate detection limits. Samples in a liquid give
very high background intensity due to the scattering of primary radiation,
and are usually dried prior to analysis. Giauque (personal communication),
however, performed analyses on liquid shale oil samples by using incoherent
scattered radiation, corrected for matrix absorption, as an internal standard.
Physical methods of drying such as evaporation by heat lamp onto a Mylar film,
absorption into spectroscopic powders with subsequent pelletizatlon, or nebu-
lizatlon onto the target substrate all have problems in ensuring even distri-
butions. The evaporation technique produces surface gradients of elemental
44
-------
TABLE 12. MATRIX ABSORPTION CORRECTION TECHNIQUES FOR XRF
itasaasSBsaBsissataaisiiaosss:asssaflasaaaiiiasaxsaaaaaaaaaaa)a 3 ssssaaaaaaaaaaaa
Technique Disadvantages
Matrix match standard
or internal stanard
Standard additions
Inter-element correction
factors (k^, j)
Modified inter-element
correction factors
Requires knowledge of the matrix composition
Excessive cosy
Contamination
Excessive cost
Contamination
Requires knowledge of the matrix composition
Requires qualitative knowledge of matrix
composition
Large cost of running n+2 standards (n°
interferences) for every sample
Transmission and fluorescence measurement
required for each sample which necessitates
more computer core memory and more expense
Direct determination of
mass absorption
coefficients
Scattered radiation
techniques
a. Rayleigh scattering
b. Compton scattering
c. True background
d. Light element
technique
aaaaaaassaaasaaaaaaasaaaaaaasaaaaaaaaiiaaaaaasasssaasaaaaaaaaaasaaaaasaaassaas
Limitations listed in Table 10
distribution while the other two methods yield poor residue recovery. The most
successful method is a chemical preparation involving formation of water-
insoluble metallic chelates and subsequent filtration. This technique is: (a)
t^me-consuming in terms of sample preparation; (b) subject to the contamination
potential associated with any preconcentrat1on technique; and (c) restricted to
a few elements (those precipitated with high efficiency by the chelating agents
such as the dlthlocarbamates). Alternate techniques have involved collection
on filters impregnated with ion-exchange resin, but again these techniques are
dependent on some prior knowledge of matrix to ensure optimum recovery of the
desired elements. Clearly, the preparation Involved in obtaining a represen-
tative homogeneous dried aliquot of a liquid sample involves considerably more
time and effort than solid sample preparation techniques.
45
-------
SolIds
Homogenelty--
Obtaining good results for solid samples 1s dependent on minimizing the
variability of matrix effects and eliminating part1cle-s1ze effects which
result from a non-homogeneous sample. The particle size that can be tolerated
depends on the measured fluorescent wavelength because the absorption coeffic-
ients Increase at the rate of approximately X3. If the x-ray beam passes
through nioerous grains before It 1s seriously attenuated, then the effect o?
the different grain sizes is averaged and the sample 1s effectively uniform.
If the x-ray beam Is absorbed In passing only a few grains, the sample Is not
uniform, and the fluorescent Intensity will depend on the part1cle-s1z® distri-
bution of each mineral component 1n a complex way.
Preparation for Measureaent--
Once In powder form, the sample can be pressed Into pellets, fused with
Hthlua borate, or slurried and filtered onto a MillIpore-type filter. A thin
uniform target sample results from filtration, but an acrylic-spray coating 1s
required to prevent flaking. Pressed pellets are homogeneous but still have
the problem of matrix effects. Correction 1s achieved using one of the methods
discussed above. When pressed pellets are used, measurements of both element
peak and background peak are required for each sample.
Fusion eliminates particle size effects and minimize- the variability of
matrix absorption by diluting the sample with flux. The fluxes commonly
employed are sodium borate, lithium borate or lithium tetraborate (which are
strong x-ray absorbers). These fluxes make the absorption from matrix sub-
stltuents negligible by comparison. The optimum dilution ratic depends on the
required detection limits and the dilution required to mitigate absorption by
matrix elements.
Loss of volatile components (As, Hg, Se) 1s a serious problem when uti-
lising fusion as the sample preparation procedure. Pell a, Lorber and Helnrich^
also experienced difficulty determining Cu, which they attributed to partla1
reduction of the Cu followed by binding to the Pt crucible. This problem can
be alleviated by using graphite crucibles or preventing reduction of Cu bv
addition of a suitable oxidant. Norrish and Hutton^®, utilizing fusion to
analyze for major elements, added the heavy absorber lanthanum oxide (to main-
tain constant detection limits) because 1t reduces background and fluorescent
intensities equally. If 1t is not used, the fluorescent intensity will de-
crease linearly with dilution while the background intensity remains constant.
For the case of trace metals, some workers recommend omission of a heavy
absorber since the La lowers the detection limits for trace element determi-
nations due to x-ray line Interferences between some of the analyte x-ray lines
and the lines of the absorber. Of course, corrections for absorption must, then
be made. Uso of more than one oxide absorber can be utilized to avoid x-ray
line interference but this will substantially Increase the sample preparation
time. 9ecause of the uniform matrix achieved with fusion, a background no
longer needs to be measured for each sample. One determination of the back-
ground value can be made using the flux material containing non of the element
46
-------
of interest and using the measured peak Intensities to correct for the matrix
of the material.
A technique similar to fusion Is to dilute the solid sample with sulfur
powder. Sulfur 1s chosen because 1t strongly absorbs x-rays, Is available as
a fine powder, does not yield x-rays in the energy range of priority pollutant
metals, does not give rise to x-rays of sufficient energy to impose on the
count-rate limitation of tne system, and Is self-binding for the preparation
of discs. Regardless of the binding material that samples and standards are
mixed with, it 1s desirable to prepare them as tiiin samples to minimize matrix
absorption and absorption corrections. Specimens should not exceed the crit-
ical thickness for Che x-ray energies of interest so that matrix enhancement
and absorption effects are minimized, count-rate limitations are not reached
on the x-ray detection system, and the Intensity and total system geometry do
net significantly vary with respect to the standards. Specimens must be thick
enough, however, that favorable counting statistics can be obtained in a rela-
tively short analysis time.
The benefits of fusion are the production of samples which are more stable
than pressed powders, reduction of matrix effects, and the elimination of back-
ground determination on every sample. Despite these advantages, Pella, Lorber,
and He1nrich(79) conclude that digital computation of x-ray absorption effects
and correction for x-ray line Interferences on pressed powder pellet samples
is a superior technique to fusion with a heavy absorber especially when dealing
with complex variable matrices and determining many elements. Fusion requires
increased sample preparation time, the inconvenience of determining the cor-
rect mixture of heavy absorbers and correct dilution ratio, and yields lower
detection limits due to sample dilution.
COSTS
XRF requires a larger capital investment than ICP, but supply costs are
lower. The main additional cost Is due to required access to a large minicom-
puter for sophisticated data reduction and correction procedures.
Capital Investment
!i75-85K (Energy Dispersive) add $15-20K for sampler (30 positions)
195K (Wavelength Dispersive, sequential)
N125K (Wavelength Dispersive - more intensity)
11350K (Wavelength Dispersive, Simultaneous, Deluxe)
There are several different commercial manufacturers who produce high
quality systems capable of trace level analysis:
• Philips - Wavelength Dispersive and Energy Dispersive
• Kevex - Energy Dispersive
Expendables
$150/450 samples - 33$ per sample based on a single questionnaire response
which provided data on this area.
47
-------
Sample Throughput
The rate-Hm1t1ng steps are sample preparation and the acquisition of
x-ray spectral data.
Operator Skill
A highly skilled operator it. mandatory because of the large amount of data
interpretation which is Involved. Only an experienced operator will be aware
of the potential problems of spectral interferences, matrix absorption effects,
and background Interferences that will occur when analyzing haza—'ous waste
samples. This awareness 1s a prerequisite to application of app "Hate correction
techniques.
Commercial Services
$5-$100/sample depending on volume of samples and complexity of analyses.
No commercial services offer determination of the 13 elements. The availability
of commercial services is much more limited than ICP since XRF has traditionally
not been used for trace level determinations. Its principal applications are
major element determinations, aerosol analysis, and basic research.
48
-------
SECTION 7
INSTRUMENTAL NEUTRON ACTIVATION ANALYSIS (INAA)
THEORY
Instrumental neutron activation analysis consists of detection of radio-
active nuclei produced when a sample 1s exposed to a flux of neutrons. Upon
bombardment of a sample, at least one Isotope of each element can usually be
transformed into a radioactive Isotope. The energy content of the generated
radionuclide is increased by an amount equal to the neutron binding energy plus
the kinetic energy of the incoming neutron. Characterization and measurement
of the energy of the emitted gamma rays yields qualitative and quantitative
analysis it the sample.
A n,y g B q
(Target Isotope} (Radioactive Isotope) (Stable Nuclei)
Formation of the radioactive isotope B is dependent on the neutron flux
density bombarding the sample (d), the number of ta-get nuclei present (N/\),
and the reaction cross section (R). The rate of change in the number of
radioactive nuclei is equal to the rate of their formation minus the rate of
their decay (equation 14).
Integrating over the bombardment time and taking Into account the delay
time prior to analysis, the relationship between activity and element concen-
tration (equation 15) Is given by:
dNo |>u
-J- - - XB K,
(14)
where
Dn " activity at the beginning of a measurement
N = avogadro's number
W = element weight
49
-------
Fa * fractional abundance target Isotope A
R » reaction cross section
4 * neutron flux
= atomic weight of isotope A
¦ decay constant of radionuclide
tfc ¦ length of bombardment time
t
-------
INAA resolution has Improved considerably since Ge(L1) and more recently
Intrinsic !ie detectors have replaced Nal(Tl) scintillation detectors. Ge(L1)
detectors can distinguish between gamma rays with an energy difference as
little as one keV. Elemental analysis of complicated matrices can now be
achieved with INAA while radiochemical separations were ofcen required to
achieve adequate resolution using a Nal(Tl) detector. The efficiency of a
Ge(L1) detector 1s dependent on its size and shape. The smaller Ge(li) detec-
tors have lower efficiency but better resolution. In order to achieve both
adequate detection limits and resolution, a sample may be analyzed with two
different-size detectors. Limited compensation for the lower efficiency of a
smaller Ge(Li) detector can be made by Increasing the counting time irradia-
tion time or neutron flux. Maintenance costs for Ge(Li) detectors are higher
than Nal(Tl) costs because liquid N2 1s needed to keep the detector opera-
tional and more expensive electronic amplification is required to amplify the
smaller output signal.
DETECTION LIMITS
The sensitivity and detection limit for each element are dependent on
the particular radionuclide which 1s monitored. Table 13 lists all the avail-
able gamma-emitting radionuclides and half-lives for the radioisotopes pro-
duced by each of the 14 specified elements plus the reaction cross sections
and isotopic abundance of the parent Isotopes. The Ideal choice is the radio-
nuclide with the largest value for the produce F*R of the parent Isotope and
the largest value for the product F«R«xB when the radionuclide is long-lived.
TABLE 13. NUCLIDES AVAILABLE IN INAA. FOR THE SELECTED ELEMENTS
saaaassaaaaiaaaaasssasaaaasassBsaasassassBaBessssaaasssassssaTSSsaaas^ssBsiasBS
Reaction Cross Section (R) Isotopic
Radionuclide (barns) Abundance (X) Half-Life, T(1/2)
(1)
AglOS
2 + 38
51.8
1.3 x 10* y
AgllO
4.4 + 88
48.17
252 days
(2)
As™
4.4
100
26.3 hours
(3)
Bal39
0.4
71.7
83 min.
38.9 hours
Ba133
0.7
0.10
10.7 years
8,131
14.6 min.
13.5
0.11
11.7 days
(4)
BelO
-
100
No gammas
(5)
Cdl09
1.1
0.89
453 days
cam
48.7 m1n.
0.1 + 11
12.5
12.8
44.6 da*-
(continued)
51
-------
TABLE 13. (Continued)
3a3a55Saa333aa833S3SSS5SS3a33S53SS3Sa3SS53SS5Sr3S5S3?SS3:SSSr.SSSSSS3SS3S3SSS5SS
Reaction Cross Section (R) Isotopic
Radionuclide (barns) Abundance {%) Half Life, T(1/2)
CdH5
0.04
28.7
Cd
0.2 + 0.8
7.5
(6)
Cr51
Cr^5
15.8
0.36
4.4
2.36
(7)
Cu64
Cu66
4.4
2.17
69.2
30.8
(8)
Hgl9?
120 + 310
0.15
Hgl99
Hg203
Hg205
0.017 + 2
4.9
0.4
10.0
29.8
6.9
(9)
N165*
N158 via Co58
1.5
0.2
0.9
(10)
Pb205
Pb207
Pb?09
0.66
0.030
1.42
24.1
52.4
U1}
Sbl22
5 + 200
57.3
Sbl24
130
42.7
(12)
Se75
Se81
Se83
5200
0.07
0.04
0.9
49.8
49.8
9.2
Se77
21 + 64
9.0
(13)
T1204
T1206
11
0.10
29.5
70.5
(14)
Zn65
Zn69
Zn71
0.76
0.072 + 0.9
0.0081 + 0.083
48.6
18.8
0.62
3333:
3-=3-s=S33-.sas-
= = = = = = = r. = = = = == = = = = :=
============
53.4 hours
3.4 hours
2.6 hours
27.7 days
3.6 min.
12.7 hours
5.1 min.
23.8 hours
64.1 hours
42.6 min.
46.6 days
5.2 min.
2.5 hours
71 days
1.4 * 107 u
0.796 sec.
No gammas
4.2 mins.
2.72 days
20.3 m1n.
60.2 days
120 days
57.3 min.
18.5 min.
69 sec.
22.3 min.
17.4 sec.
No gammas
3.6 min.
243.8 days
13.8 hours
2.4 min.
3.97 hours
*Iostopic Abundance too low for determination.
52
-------
In the case of mult1elemental analysis, detection capacity must some-
times be sacrificed 1n order to avoid serious Interferences with the other
elements being determined and the major matrix constituents. Beryllium cannot
be analyzed by INAA because 1t forms no gamma-emitting radionuclide. Only one
gamma-emitting nuclide exists for Pb and T1, and the very low reactior cross
section for these nuclides implies poor detection limits for Pb and T1 determi-
nations by INAA.
Instrumental variables which affect the detection limits are the irradia-
tion tlra, decay time, detector efficiency and neutron flux. Detection limits
can typically be Improved by a factor of 10-10-^ if the irradiation time and
neutron flux are Increased. INAA 1s based on radioactive courts which obey
the Polsson distribution (unless the number of counts 1s large and a normal
distribution Is approximated). The definition of detection limits is typically
based on the work of Currie and, as for XRF, 1t is not clear whether a particu-
lar author 1s using the Polsson- or Normal-distribution definition.
The median detection limits for the elements 1n assorted matrices (such
as rock, coal, ?and, liver and leaves) are shown 1n Table 14. No data for Pb
and T1 were included because the high detection limits reported (>100 parts per
thousand) are outside the concentration range of interest. The detection
limits for Ba and N1 are significantly higher than any of the other elements
with values of approximately 50 and 150 ug/g, respectively. Ni has only one
Y-emitt1ng radlcnucliderand the parent isotope abundance 1s too low for accu-
rate determination. Ni58, when irradiated, will emit a proton and form Co58,
which can be used for N1 determinations. The high detection limit observed
for N1 is probably due to the low reaction cross section of the N158 qq58
reaction 1n reactors which have only a small component of fast neutrons. With
reactors having a fast/slow neutron ratio of approximately 1, the N1 detection
limit 1s about an order of magnitude better. The poor detection limU for Ba
Is probably due to severe Interference from Fe which 1s very abundant in many
types of samples. An alternative explanation for the high Ba and Ni detection
limits Is the fact that the Ba and N1 databases were dominated by values re-
potted by a commercial laboratory which may not have been striving for optimum
performance. The detection limits for the remaining elements are all under
10 ug/g.
TABLE 14. MEDIAN OF REPORTED INAA DETECTION LIMITS*
3 3 a a a 3 a 333a 3 a 3 a 3 3 S 3 3 S 3 = a 3 n 3 3 3 S3 33 ? a = 33 3a 3 2 2 3 = .3 a a 3 S 5 = = = 3 S S 3 3 S 3 a e 3 5333 3 3 = S 7 S S = 3 = s
Detection Limit Detection Limit
Element (uq/q) Element (uq/q)
Ag 2 (5) Hg 0.6 (5)
As 1 (6) N1 160 (5)
Ba 53 (7) Pb NA
Be NA Sh 0.6 (7)
Cd 8 (6) Se 0.6 (9)
Cr 3 (7) T1 NA
Cu 6 (6) Zn 5 (7)
8SI3a8333935a333338SS«S>33 JS5S353338a3S33333SSS>33338333 33S3S333 533S33&3SSS3S353S
*Parantheses contain the number of reported detection limits
NA = Not Available
53
-------
ACCURACY AND PRECISION
In theory, INAA precision should be primarily a function of counting
statistics. Other errors are minimized by matching counting geometry and
irradiation conditions for samples and standards and usinq a high resolution
Ge(Li) analyzer. Accuracy should be excellent since most interfering ele-
ments are not present in large abundance in most terrestrial samples.
The mean precision (percent RSD) and accuracy (pe.cent of true value) for
each element determined by INAA as a function of concentration is shown in
Table 15. These graphs exhibit the same trend with concentration as the pre-
cision and accuracy versus concentration data for ICP and XRF. Precision
improves with Increasing concentration while the accuracy is excellent for the
high levels but poor for some of the elements at low levels (<10 pg/g). No
data was collected for Ag, possibly because its detection limit near 1 pg/g is
comparable or larger than the level in most terrestrial materials used as
SRMs. The precision exceeds 20 percent for As at the 10-ug/g level, for Ba at
100 pg/g, for Cr at 10 y?/3» for Ni at 100 pg/g, and for Sb at <1 pg/g. The
poor N1 precision probably reflects measurements with reactors having few fast
neutrons, while Ba suffers from Fe Interference. The acccuracy becomes unac-
ceptable (exceeds arbitrary 20 pe-cent error criterion) at a level of 10 pg/g
for Cd, 10 ug/g for Cr, <1 pg/g for Hg, 10 pg/g for Ni, and 10 pg/g for Sb.
The Cu datum is not considered significant since it is based on only one piece
of data.
SOURCES OF ERROR
Analytical ep-rors in INAA occur when factors affecting the formation of
the radioactive isotope or detection of the emitted radiation differ signifi-
cantly between the sample and the standard. Formation of the radioactive
Isotope is dependent on the neutron flux density, the number of target nuclei,
and the reaction cross section. The number of carget nuclei and the reaction
cross section are intrinsic properties of the sample, and variations in the
flux density hitting samples and standards are tr.vial due to constant rotation
of the specimens within the reactor. It is primarily the detection of the
emitted gairana-rays which is a potential source of error. The accuracy of the
total photopeak count must be corrected for systematic errors arising from
spectral interferences, peak-broadening errors absorption errors and instru-
ment dead time losses. Random errors result from counting statistics and
procedural uncertainties.
Spectral Interferences
Spectral interference occurs when the gamma-ray radiation emitted by a
matrix element cannot be resolved from the radiation emitted by the element of
interest. The most common interferences result when different nuclear reac-
tions produce the same radionuclide. An example is the formation of Cr51 from
Fe^ through loss of an alpha particle which will interfere with a Cr determi-
nation. Radioactive decay catalogues are useful toolo for predicting inter-
ferences but more obscure interferences such as p1le-up peaks will be over-
looked. Pile-up peaks result when 2 Y-rays are counted coincidentally. If the
54
-------
Element
kg
NA*
NA
As
Below 1
U
105
1-10
36
97
10-100
9
NA
Ba
1-10
NA
103
10-100
39
NA
100-1000
5
102
Be
-
NA
NA
Cd
1-10
NA
58
10-100
NA
96
Cr
Below 1
29
NA
1-10
24
78
10-100
6
106
Cu
10-100
13
130
Hg
Below 1
NA
150
1-10
NA
90
10-100
NA
SO
Ni
Below 1
300
NA
1-10
136
3?
10-100
38
94
100-1000
14
104
Pb
10-100
NA
108
10C-1000
NA
88
Above 1000
NA
95
Sb
Below 1
58
64
1-10
14
160
Se
1-10
8
100
T1
-
NA
NA
Zn
10-100
10
NA
100-1000
8
NA
Above 1000
1
NA
NA = Not Available
55
-------
energy sum of these coincident peaks is in the neighborhood of the peak of
interest, an interference results. Table 15 lists the INAA spectral interfer-
ences for the elements uf interest.
Several procedures are available for attempting to eliminate spectral
interferences. Peak resolution of interfering radionuclide emissions can
sometimes be achieved using a small Ge(Li) detector with optimal resolution.
If resolution still cannot be achieved, the contribution fr )in th? interferer.t
peak may be removed by determining the phoiopeak area of the interferent at
another gamma-ray energy line, and subtracting the interferent value from the
peak of interest. Interference effects can sometimes also be minimized bv
adjustment of the irradiation time. Lowering the irradiation time may alle-
viate the interference if the element of interest was being obscured by the
high matrix activity being produced at the longer irradiation times. The
reduced detection limits which result from a decreased irradiation time must be
tolerable. If none of these correction techniques work, it may be necessary to
switch to a different energy line even if this substantially reduces the detec-
tion limits. A final, more elaborate, method of interference correction in-
volves an extension of the counting technique by using NaI(T1)-Ge(Li) coinci-
dence measur3ments. The coincidence/non-coincidence counting technique uti-
lizes both Nal(Tl) and Ge(Li) detectors operating simultaneously. Gamma-ray
emissions detected by both detectors are stored in half ot the channel memory
as coincidence events while events seen only by the Ge(Li) are stored in the
other half of the memory as non-coincidence events. This separation results
in a significant reduction of the Compton background and interference effects
which leads to greatly improved detection limits. Overall, non-coincidence
counting will increase accuracy and relative precision for mary elements by a
factor of appoximately 10 relative to simple Ge(Li) counting.
Absorption Errors
Errors caused by matrix absorption of an activated element's emitted
Y-rays is rarely a problem unless elements with large Y-absorption cross
sections are present in the sample anc* not in the standard.
Dead-Time Loss
Instrument dead time is the non-responsive period when the instrument is
accepting one signal impulse. When the signal rate exceeds a certain level,
the proportion of impulses that are recorded decreases. Dead-time correction
involves compensation based on the signal rate.
Counting Statistics Errors
In accordance with normal distributions, the error due to the statistics
of radioactive counting is given by the square root of all the counts used co
measure a particular element (VGross peak count + background count). Determi-
nation of strongly activated elements or large concentrations will yield a
large number of counts and small errors. The number of counts is also depen-
dent on the count time and detector efficiency. The dispersion in the pre-
cision can be expressed by the root-mean-square deviation. Separation of the
56
-------
TABLE 15. PRINCIPAL INTERFEflENTS IN I KM FOR THE SELECTED ELEMENTS
Radionuclide Interferents
(1) As76 (659 keV) Br®2 (554 kef), C&193 (ssa keS>,
Ge77 (563 keV), Sb122 (J64 keV)
[2) to111 (123.? teV) Eu154 (123.1'keV)
(496 keV)
Ba139 (166 keV)
CdllS (M93 kev), Ce*43 (493 ke¥),
Pd1^ (498 tcel'), U?15 (n,fission},
ru103 {497 keVj
Ce^9 (165 ke¥). to£39 (156 keVJ,
Ce^ (l66 keVj Mg27 n7i fceV},
Cfiif7Tn <163 k«V), C«^X"(I69J,
!j235 (n,fission;, Ba"9 (156 keV)
(3) Cr51 (320 keV) TaiS2 (100+222=322 fceV),
^09 {"V310 kev), (317 ice/),
Ndl#/ (319 keVJ, Ir*94* (^320 kel'h
S>tW (v32o keV), Qs1^ (3Z2 keV).
Ru9? (324 kev), Fe§4 {n> Qj £rSl
LtA>7 {113+208=321 keV)
(4), Cu64 (511 fceV) M^5 >*.&), Tem (508 VeV),
Mo**" bS10 keV}, Co53 (511 keV),
Zr6S (511 keV), M? (511 keV I,
ttr'3 (511 keV), (Cr95 (514 keV),
Cd^7 (511 fceV), Sr36 (5H keV}.
Se83 (-V520 keV). Na24 (511 from pair
prcductidflj.
(5) vfa Co58 (310 keV) Ca47 (807 keV), Ho1®611 (810 keV),
La140 (815 keV)
(6) Sbl22 (564 keV} As™ (559. 563 keV*), Cs134 (563 . 559 keVJ
Sb"4 (1592 kef}
(7) Se75 (136 keV} Hf1S3 (137 keV)
(8) Zr>65 (ni6 keV) 7e12i™ ^UDD keV), Er1^ (^100
C€143 (%noo keV), Kr79 (1119 keV),
Zn71m ^noo keVK Sc46 (U20
£C (H15 keV)
:3tf35;s55;
* OiV5.y underHriecJ values are 5ignifleant in high precision jDeaturements*
5?
-------
counting errors yields a residual dispersion resulting from procedural uncer-
tainty. This uncertainty (which Is smaller than 1} at the Lawrsnce Berkeley
Laboratory) arises fraa small residual random errors due to variations 1n
neutron flux, variations In specimen geometry, small dead-time errors in the,
analyzer system and random errors in sampling and weighing.
CORRECTION METHODS
For experienced, high-quality laboratories (e.g., the Lawrence Berkeley
Laboratory) precision is wel1-reDresented by counting statistics and other
errors have betn held to lew values by intensive investigation of the causes of
these errors. The accuracy of the analysis of an unknown sample can be checked
by determining a particular element several times utilizing a different-energy
gamma ray each time, by measuring the same gamma ray by different detectors,
or re-measuring the gamma ray at a later time when the Compton-scattering
background has changed.
Use of several irradiations for different periods and counting after dif-
ferent decay periods eliminates many of the potential sources of error. Short
Irradiation times allow the analyst to obtain data for short-livsd isotopes
without generating high specific activity from longer lived isotopes (with a
potential for large (lead-time corrections). Alternate decay periods make it
easy to resolve peaks of similar energy but different half-lives. Adequate
computer software is essential for performing these computations.
SAMPLE PREPARATION
INAA involves no elaborate sample preparation procedure. Solids are
ground and a representative aliquot is mixed with cellulose and pressed Into
a standard size pill. Care should be exercised to ensure that the analyzed
aliquot represents the bulk sample. Analysis of several aliquots can serve to
demonstrate the homogeneity of a sample material. Use of a standard pill size
ensures maintaining the solid angle from the fill to the detector which relates
directly to the accuracy of the method. Cr contamination at the level of 0.$
ug/g sometimes results from the tool-steel dies used to prepare the pellets.
Solutions can be analyzed after freeze drying and pelletizing the resulting
powder.
COSTS
Capital Investment
$500K (requires permit)
$135 and up depending on
required length of
irradiation, sample
size, and neutron flux
$30K and up
*or access to irradiation source
Nuclear Reactor*
Contnercial Irradiation Cost (for short-lived
isotopes the detector must be at the
irradiation site)
Gaitma-Ray Spectrometer & Detector
58
-------
Expendables
Expendable costs are largely dependent on number of samples analyzed.
Based on a single questionnaire response, costs were estimated to be $l-$2 per
sample for expendables and approximately $2-$3 per sample for labor.
Sample Throughput
The rate-limiting step is the time required before interfering radiation
has decayed sufficiently to allow long-lived isotopes to be counted. This can
be as long as 30 days for the analysis of long half-life isotopes such as Cr^l.
Throughput can also depend on the cross sections of matrix Interferents as wel1
«s the half-life of the element of interest. If matrix elements become highly
radioactive, 1t may be difficult to detect the small amount of radiation emit-
ted by a trace element. In this case, provided the interferents have shorter
half-lives than the elements of interest, it is necessary to wait until the
interfering radiation has decayed away before counting the sample.
Operator Skill
As in XRF, a highly skilled operator with a large amount of experience is
required for successful analysis by INAA because of the complexity of Interfer-
ences tftich occur.
Conrerclal Services
Typical cost for single sample multielemental analysis is $400 with de-
livery times of S-6 weekr after receipt of sample. Availability of commercial
services is quite limited, because of the difficulties 1n acquiring access to a
nuclear reactor. Services which use university reactors are available, but
typically only on a research basis and not for routine analyses. It is pos-
sible to perform INAA using a radioactive element as a neutron source (such as
Cf25*)» but the flux obtained from such a source is typically not adequate for
trace level work.
59
-------
SECTION 8
COMPARISON
INTRODUCTION
Comparison of the three techniques Is based on the following parameters:
• Detection Limits
• Accuracy and Precision
• Interferences
• Flexibility
• Sample Preparation
• Cost
• Throughput
• Availability
Table 17 provides an overall comparison of the three techniques by ranking them
for each of these factors.
TABLE 17. RANGING COMPARISON Of ICP, XRF AND 1NAA
asaaasaasazzsasaassaaaasaaaassaasssSBs:assssssssssssasasssassaassaaaaassasssaaa
Relative Ranking
Characteristic (Best to Worst)
(1)
Aapid analysis
XRF>ICP>INAA
(2)
Detection Limits
ICP, INAA>XRF
(3)
Precision and accuracy
1CP>INAA,XRF
(4)
(5)
Ease of error corrections
INAA, ICP> XRF
Low cost
ICP>XRF>>.'NAA
(6)
Flexibility
ICP, INAA>XRF
(7)
Availabi1ity
ICP>XRF>INAA
(8)
Ease ot sample preparation
INAA>/RF>ICP
DETECTION LIMITS
Table 18 summarizes the median
function of instrumental technique,
tion limit units into (ug/g) solids
tlon of 100 mg of solid sample with
detection limits for each element as a
Conversion of the aqueous mg/1 ICP detec-
for this table was made by assuming diges-
10 ml of acid. Division by 10 will convert
60
-------
TABLE 18. COMPARISON OF MEDIAN DETECTION LIMITS FOR ICP, XRF ANO INAA
iJisaissaaiassssiaisBaaaixBsaaaassiaisstaiaBaBxsaiaBvastavsasssasssssaflaBasaassr
Method
Element ICP (119/9) XRF Ug/g) INAA (ug/ql
Ag
0.3
—
1
As
3
0.9
0.2
Ba
0.05
10
53
Be
0.05
--
--
Cd
0.2
5
15
Cr
0.3
1
2
Cu
0.2
1
6
Hg
3
2
0.3
Nt
0.8
1
160*
®b
2
2
--
Sb
3
o.rw
Se
2
13
0.3
T1
2
--
--
Zn
2
3
5
a aaaaasasaBsasaassaasasaaasaaaaaaaasaaassaasaaaasaasaaaasaaassasaaaaasnaasaas:s
* Results for N1 probably biased by database (see INAA section).
eg/I to ug/g which enables direct comparison of ICP detection limits with those
obtained for XRF and INAA. The detection limits of ICP and INAA is slightly
superior to XRF.
ICP 1s the only technique capable of determining all 14 specified ele-
ments. Beryllium cannot be determined by INAA {because its radionuclide emits
no gamaa rays) nor conventional by XRF (because the long wavelengths of the
emitted x-rays cannot be dispersed). Pb and T1 are too weakly activated to be
successfully determined by INAA, except at very-high concentrations. The lack
of a substantial database for many trace elements by XRF reflects its common
use as a major-element technique.
ACCURACY ANO PRECISION
Comparisons are provided 1n Tables 19 and 20 of the reported pn»c1s1on
and accuracy for the 14 specified elements by the three techniques. Only data
collected on rock, sediment, soil, sludge, coal, oil, steel or pottery were
considered since these sample types most closely approximate hazardous waste
matrices. The comparisons 1n Tables 19 and 20 were made by calculating the
perctntages of "acceptable" precision and accuracy data each technique pro-
duced. For this comparison, acceptable precision was arbitrarily defined as
a re'latfve standard deviation rot exceeding 20 percent and acceptable accuracy
was arbitrarily defined as a percentage of the true value between 80 percent
and 120 percent. For some of the elements, Insufficient data were available
for nil three techniques to allow a valid comparison. This 1s the case for Aq,
Be, Cd, Cu, Hg, Pb, Sb, Se, and T1 precision data. There is only sufficient
precision data by all 3 techniques for comparison of As, Ba, Cr, N1 and Zn.
For Ea and Zn, the differences In the percentages of acceptable precision among
61
-------
TABLE 19. PERCENTAGE OF CASES WITH ACCEPTABLE PRECISION*
saaasassaasssaaaissaaaasBassssssssxassaaasasssLissxsssasaazsssasss&ssssassssass
Percentage Acceptable
Element ICP XRF INAA
Ag
80 (5)**
NA
NA
As
86 (7)
60 (5)
54 (11)
Ba
100 (12)
92 (12)
92 (35)
Be
100 (7)
NA
NA
Cd
95 (27)
0 (1)
NA
Cr
100 (28)
80 (34)
77 (26)
Cu
88 (49)
70 (17)
100 (1)
Hg
NA
NA
NA
N1
84 (43)
81 (38)
47 (15)
Pb
97 v 30)
82 (11)
NA
Sb
100 (2)
NA
54 (13)
Se
NA
100 (1)
100 (2)
Tl
NA
NA
NA
Zn
97 (40)
100 (20)
92 (12)
asassaatsasasasssassaaaiassaaaxaiiiaaasasaasaaaflaaaaaaasaaaaasiaasaaaasaaaiaasa
~Acceptable precision defined arbitrarily as a 20 percent relative standard
deviation.
~~Parentheses contain the number of cases.
NA 3 Not Available
the three techniques are Insignificant. ICP provided more high-precision
results than the other two techniques for most of the elements.
Inspection of Table 20 Indicates 1nsuff1cent data for Ag, Be, Hg, Sb, Se
and Tl. Of the eight remaining elements, ICP gave more high-recovery data than
the other two techniques for Cd, Cr, Cu and Ml. There 1s Insufficient data for
comparison of As by XRF and INAA. INAA gave better results for Ba, and XRF
gave better results for Zn. The case of Pb Is somewhat misleading as 1t appears
that INAA gave comparable results to XRF. This observation would be disturbing
since INAA 1s known to be a poor trace technique for Pb. Closer Inspection
revealed that the INAA Pb results are based on only four samples with extremely
high levels of Pb. A suimary of the results given 1n Table 21 clearly Indicates
that, overall, higher precision and accuracy for a majority of the elements are
obtained with ICP.
INTERFERENCES
The analysis errors Involved In any of the three techniques can be at-
tributed to specific, causes. Random errors which affect the precision can be
caused by instrumental drift or may arise from counting errors for INAA and
XRF. Systematic errors which affect accuracy result from spectral interfer-
ences occurlng In the sample matrix as well as remaining nonspectral Inter-
ferences such as matrix absorption or background Interference. Samole prep-
aration errors may be either systematic or random.
62
-------
TABLE
20. PERCENTAGE OF
CASES WITH ACCEPTABLE ACCURACY*
isasaaaaaasaaasss
laasaaaaaaKaaaasaasi
S33SaS088a333 3SaS33S533SSSa3 3S3 3a
Percentage Acceptable
ssaasaaaaa
ement
ICP
XRF
INAA
Ag
69 (16)**
NA
NA
As
85 (13}
100 (2)
100 (2)
Ba
89 36)
78 (71)
100 (16)
Be
53 (15)
NA
NA
Cd
58 (26)
25 (4)
50 (4)
Cr
68 (54)
64 (79)
53 (18)
Cu
94 (76)
62 (39)
0 (1)
Hg
NA
NA
80 (5)
Ni
80 (47)
78 (90)
38 13)
Pb
58 (63)
70 (34)
75 (4)
Sb
NA
NA
14 (7)
Se
NA
100 (1)
100 (1)
T1
NA
NA
NA
Zn
81 (43)
97 (49)
88 (17)
asisataaaassasssaasjassasiassasasasasassaassaaasassaa^aaaaasaassaasaassaasasass
•Acceptable accuracy defined arbitrarily as recovery between 80-120%.
**Parentheses contain the number of cases.
NA = Not Available
Table 22 indicates t«e relative importance of these error sources for
the three techniques. For both INAA and XRF, ths counting errors are least
significant and never amount to more than 1 percent. For XRF, sample prep-
aration is the most Important source of error because it is directly related
to the non-spectral errors which are a major problem in XRF. Unless sample
preparation 1s adequate, good data Mill never be generated no matter how elab-
orate the non-spectral correction procedures are. Obviously, non-spectral
errors rank second In importance followed by spectral Interference errors and
then Instrumental errors. The different types of spectral interferences are
well documented and can usually be anticipated and corrected. Adequate cor-
rection for spectral and nor,-spectral errors is completely dependent on the
operator's skill and experience. Instrumental error results in a small re-
sidual background. This error can be correi*.ed by plotting the spectral back-
ground as a function of the Compton-scattereu radiation and discerning the
value of the y-intercept. For INAA, both instrumental and sample-preparation
errors are minimal. The major source of error is spectral interference ?-d
again the major requirement for adequate correction Is an experienced operator
(skilled in making comparative cross-checks for an element by measuring dif-
ferent gamma-ray emissions).
The major source of ernjr for ICP 1s sample preparation because of the
potential for contamination or loss of analyte during this step. Spectral
interference is the next most impo-tant source of error again because correc-
tion depends on the skill and experience of the operator in recognizing these
63
-------
TABLE 21. RANKING OF TECHNIQUES BY ELEMENT
ssaaasasaaaasassssaaaasasaasaasssaassassa:sssaaaasssss^sssssasass:::^;::^
Element Precision Accuracy
Ag No data for INAA, XRF No data for INAA, XRF
As 1CP>1NAA, XRF Insufficient data for
INAA, XRF
Ba ICP>XRF, INAA INAA>ICP>XftF
Be Mo data for INAA, XRF No data for INAA, XRF
Cd No dat* for XRF ICP>INAA>XRF
Insufficient data for INAA
Cr 1CP>1NAA, XRF ICP>XRF>INAA
Cu ICP>XRF ICP>XRF
Insufficient data for INAA Insufficient data for IHAA
Hg No data No data for ICP, XRF
Ni ICP>XRF>INAA 3CP, XRF>INAA
Pb ICP>XRF XRF>ICP
No data for INAA Insufficient data for INAA
Sb ICP>INAA No data for ICP> XRF
No data for XRF
Se Insufficient data Insufficient data
T"! No data No data
Zn ICP, XRF>INAA XRF>INAA>ICP
::S33S3;a:a3J33;2r;2=S3»3Sa5lISC=33333333a333S333 3S3:35Saa3*333333333 333=;33SSM
TABLE 22. RANKING OF ERROR SOURCES FOR THE TECHNIQUES
SS333333SdaS33aS33 55Saa3a333 33 3333£SS3 33 3:aS3333 333333SSSS33383333333:338339333
Sample Non-
hcfi Instrumental Counting Spectral Spectral
ICP I 3 4 2
INAA 3 4 5 2 1
XRF 1 4-5 5 2 3
3 333333aSd3Saa3333S3S33?3333a3333a33«:a3333S3333a333933333a33933334333333333S8*
I = High (very important)
5 * Low (trivial)
64
-------
errors. Instrumental drift Is usually not a serious problem 1f re-calibratlon
is performed periodically, and computer corrections for non-spectral background
interferences are usually quite adequate.
Table 23 outlines the various combinations of analytical measurements
•which can be utilized to separate these five sources of error in order to
quantify each source Individually. Although insufficient information was found
in the literature to be able to achieve this type of error analysis with the
collected data, the outline in Table 23 could be applied to an experimental
comparison of the three methodologies. Separation of all the different sources
of error can be achieved except for separation of the spectral and non-spectral
interferences.
TABLE 23. ANALYTICAL DEFINITIONS OF RANDOM AND SYSTEMATIC ERRORS
saa33as-339sa38aaaa33saa8a<>saasaaii3as5«53aasa*8s8as3s3533s5sss33333s33saaaa^aa
Error Analytical Definition
1. Instrumental and counting error
2. Instrumental and counting and
random sample prep error
3. Random sample prep
4. Counting error (INAA and XRF)
5. Instrumental Error (INAA and XRF)
6. Random sample prep and systematic
sample prep and spectral and non-
spectral error
7. Systematic sample prep and
spectral and non-spectral error
8. Spectral and non-spectral error
9. Systematic sample prep error
3338333333 3S3.1 3333333333333« 33a 33333333333
Precision of replicate analyses
on one sample
Precision of one analysis on
replicate samples
(2) minus (1)
V Total Counts
(1) minus (4)
Accuracy of replicate determina-
tions on one sample
Accuracy of one determination on
replicate samples
Accuracy of one determination on
replicate standard reference
materials by two different methods
(e.g., ICP and AAS)
(7) minus (8)
33 38333 883 3 333333333 3393333333333333
FLEXIBILITY
ICP also has advantages 1n terms of flexibility. With a scanning mono-
chromator, 1t Is easy to switch to a different spectral line without the rigid
time constraints Involved in IKAA. For the case of XRF, there is little choice
65
-------
in the fluorescent energy which can be monitored for a particular element since
only a limited number of energies are available.
SAMPLE PREPARATION
The cne area where ICP lags behind INAA and XRF is in ease of sample prep-
aration. Since conventional ICP is not suited for the direct analysis of sol-
ids, a digestion step is needed, which, adds expense and potential for errors.
However, sample digestion does serve to homogenize the aliquot before the
analysis.
COST
ICP is the least costly of the three techniques, both in terms of capital
investment and operating costs. In addition, conmercial services for ICP are
low in cost compared to XRF or INAA. The cost of operating an XRF system is
comparable to ICP although more extensive computer facilities are required.
The expense of operating an INAA system is on the order of 5 to 10 times that
of ICP or XRF and even conmercial servics fees are 5 to 10 times higher.
Throughput
XRF is far superior to ICP and INAA in its ability to generate a rapid
semi-quantitative analysis scan of a sample in less than 5 minutes. The sample
preparation time for ICP takes well over an hour while INAA requires weeks to
allow sample activity decay for certain elements. XRF could, therefore, have
utility as a preliminary screening technique for hazardous waste samples of
totally unknown composition.
Availability
There are currently many more manufacturers of ICP systems and greater
availability of this commercial service than for XRF and, especially, for
INAA. In the case of XRF or INAA commercial services, the higher skill levels
required for accurate interference corrections at trace levels make the actual
choice cf conmercial laboratories important in achieving accurate results. For
ICP this is less crucial since most conmercial software incorporates appro-
priate correct ions.
Conclusions
After considering the detection limits, precision, accuracy, errors and
correction procedures, e:.pense, flexibility, availability and samole prepa-
ration, ICP appears to be most amenable of the three techniques to analysis of
hazardous waste samples while XRF may be useful as a preliminary screening
technique. There is an Insufficient database in the literature for some of the
specified elements. A literature review suffers from:
a. A lack of comparable databases for individual techniques with respect
to sample types, definition of detection limits, and elements.
6b
-------
b. An Inherent bias exists that published data are typically the best
available and not indicative of what may be observed In routine
applications.
Because of these problems, a fully satisfactory comparison of the methods
must be based on analyses of Identical wastes by the techniques.
67
-------
REFERENCES
1. Abercrombie, F. N. and Cruz, R. B., "Determination of Trace Inorganic
Toxic Substances by Inductively Coupled Plasma - Atomic Emission Spectros-
copy," ACS Symposium Series No. 94, 174th Symposium Meeting of ACS,
Monitorinq Toxic Substances, Chicago, Illinois, August 31, 1977, pp.
113-135 (1979). Precision and accuracy data for Ba, Zn, N1, Cu, Be, Sr,
and V in standard rocks. Description of operating principles and types of
nebulizers. Discussion of sample dissolution procedures and interference
corrections for spectral interferences.
2. Abercrombie, F. N., Silvester, M. D., and Cruz, R. B., "Simultaneous
Multielement Analysis of Biologically Related Samples with RF-ICP," ACS
Advances in Chemistry No. 172, Ultratrace Metal Analysis and Biologically
Science and Environment, pp. 10-26 (1979). Precision and accuracy data
for various trace and major elements in biological matrices such as
orchard leaves, bovine liver, grains, and fish. HNO3/HCIO4 reflux diges-
tion used on all samples.
3. Ackermann, F., Bergmann, H., and Schleichert, U., "On the Reliability of
Trace Metal Analyses: Results of Intercomparison Analyses of a River
Sediment and Estuarine Sediment," Fresenius Z. Anal. Chem., 296, pp. 270-
276, (1976). Determination of all 14 EPA-regulated elements, with the
exception of Be and T1, in river and estuarine sediment. Interlaboratory
evaluation of AAS vs. INAA indicated more difficulties with analysis by
AAS for the estuarine sediment. Enumeration of sources of INAA systematic
error.
4. Ajar, R. M., Calager, P. D., and Davison, A. L., "Multielement Analysis
with an Inductively Coupled Plasma/Optical Emission System," Mierican
Laboratory (March 1976). Description and operation of plasma"! Instru-
mental detection limits of 2 x standard deviation of background given for
all priority pollutant metals except As and T1. Cu precision data as a
function of concentration indicates substantial reduction in percentage
RSD at the 0.1 ppm level.
5. Asaro, F., "Analysis of ARCH0-1," Lawrence Berkeley Laboratory Publica-
tion, pp. 1-16, (November 9, 1976)" Analysis of basalts by INAA, soft XRF,
hard XRF, and wet chemical methods for major elements. Trace element
results and precision were reported for the method giving the best values.
The INAA calibration procedure and errors of measurement are also
discussed.
58
-------
6. Asaro, F., "Analyst. of ARHCO-l," Lawrence Berkeley Laboratory Publica-
tion, pp. 1-15, (November 9, 1976)1 Analysis of basalt by INAA and soft
and hard x-rays. Precision data ger.eratad from 1-7 replicate analyses on
duplicates. Discussion of INAA sources of error.
7. Asaro, F., "Applied Gamma-Ray Spectrometry and Neutron Activation
Analysis," Colloquium Spectroscopy cum Internationale, International
Conference on Atomic Spectroscopy, Praha 1977, pp. 413-426, (1977).
Discusses factors affecting precision and accuracy. None of the 14 EPA
regulated metals are represented in the precision ctata.
8. Asaro, F., Michel, H. V., and Burger, R. L., "Chemical Source Groups in
Ecuadorian Obsidian," Lawrence Berkeley Laboratory Report 13247, pp. 1-27,
(1981). Analysis of archaelogical obsidian by INAA using XRF as a
preliminary screening technique. Precision represented by counting
statistics since other random errors and systematic errors were held
below 1%. Major systematic error is uncertainty of the standard pottery
used for calibration,
9. Bazan, F. and Bonner, N. A., "Absorption Corrections for X-Ray Fluores-
cence Analysis of Environmental Sfrnples," Advances in X-Ray Analysis, Vol.
19, pp. 381-390, (1976). Demonstration of a linear relationship between
the absorption coefficient of an element and the ratio of the Compton to
Rayleigh scattering by the sample containing the element. This eliminates
the need to measure the absorption coefficient for each sample in a set of
samples with similar matrices.
10. Billiet, J., Dams, R., and Hoste, J., "Multielement Thin Film Standards
for XRF Analysis," .X-Ray Spectrometry, Vol. 9, No. 4, pp. 206-211, (1980).
Kethodoloqy for preparation of tMrwilm multielemental standards using a
tracer to accurately derive concentrations. Calculated elemental concen-
trations were validated by ICP and INAA analysis. These standards are also
useful to check the empirical correction factors for x-ray absorption.
XRF analysis of these polymer standards generated precision data and
indicated an accuracy problem for As, possibly due to volatilization during
preparation or irradiation.
11. Brady, F. P. and Cahill, T. A., "Development of X-Ray Fluorescence
Analysis and Applications," Final Report to the NSF Division of Research
Applied to National Needs, pp. 1-116, (April 30, 1973). Discussion of
development and current methodology for XRF and ion-excited x-ray
emission. Focuses on instrumentation, calibration, sample preparation
techniques, spectral corrections, and matrix absorption corrections.
Specific application programs are also discussed,
12. Boumans, P. W. J. M., Bastings, L. C., de Boer, F. J., and Van Kollenburg,
L. W. J., "ICP-Atomic Emission Spectroscopy as a Tool for Flexible Single
Element Analysis of Non-Routine Samples," Fresenius Z. Anal. Chem., 291,
pp. 10-19 (1978). Examines use of ICP for single-element determinations
in non-routine matrices. Gives detection limits for 44 elements under the
compromise conditions used for simultaneous multielemental analyses and
Investigates matrix salt effects.
69
-------
13. Boumans, P. W. J. M., "Inductively Coupled Plasmas (ICP): State-of-the-
Art in Research and Routine Analysis," Colloquium Spectroscopicum
Internationale and International Conference on Atomic Spectroscopy, Praha
1977, pp. 7-25 (1977)» Description ot instrumentation and comparison of
nebulizers. Discussion of physical and spectral interferences. Concludes
that the precision is limited by nebulizer stability. Review of ICP
applications.
14. Bradbury, H., "Biological Sample Analysis with ICAP Spectroscopy,"
Jarrell-Ash Plasma Newsletter, Vol. 1, Ho. 2, pp. 3-4 (April 1978).
Presentation ot accuracy data for Cu, Zn; and other major and trace
elements in NBS biological samples such as orchard leaves, tomato leaves,
and alfalfa.
15. Brewer, P. G., Spencer, D. W., and Robertson, D. E., "Trace Element
Profiles from the Geosecs-II Test Station in the Sargasso Sea," Earth and
Planet. Sci. Lett., 16, pp. 111-116, ( 1572). Trace element data on ocean
water profiles. 7n determined by solvent extraction AA and INAA at
different laboratories.
16. Brown, G. C., Hughes, D. J., and Eason, J., "New XRF Data Retrieval
Techniques and Their Application to USGS StandarJ Rocks," Chem. Geol., 11,
pp. 223-229, (1973). XRF determination of major and trace~elements in
USGS standard rocks utilizing pressed pellets made from homogenized rock
powder. Accuracy data presented for the major and trace elements.
17. Butler, C. C., Kniseley, R. N., and Fassel, V. A., "Inductively Coupled
Plasma - Optical Emission Spectrometry: Application to the Determination
of Alloying and Impurity Elements in Low and High Alloy Steels," Anal.
Chem., Vol. 47, No. 6, pp. 825-829 (May 1975). Determination of 12 metals
including Cr, Cu, Ni and Pb in steel using HCI/HNO3 digestion. Gives
detection limits based on 2x standard deviation of background for water
and 0.5% Fe solution, and reported no significant deterioration due to the
Fe matrix. All values determined for NBS steels were accurate despite Fe
matrices ranging from <1% to >99%.
18. Carr, C. D. and Borst, J. E., "Applications of an Inductively Coupled
Plasma Spectrometer to the Analysis of Difficult Samples," Applied
Research Laboratories Pub!ication, pp. 1-7. Discussion of nebulization
difficulties when analyzing samples with high salt concentrations or
unusual physical properties. Suggests matrix matching or internal
standardization. Precision and accuracy data presented for spiked brines
and phosphoric acid.
19. Charalambous, G. and Bruckner, k. "Technical Application Note No. 1,
Analysis of Metallic Ions and brewing Materials, Wort, Beer, and Wine by
Inductively-Coupled Argon Plasma Spectroscopy," Jarrell-Ash Plasma
Newsletter, Vol. 1, No. 4, pp. 4-9 (October \978)"^ Multielemental deter-
mination of metallic ions in brewing materials, wort and beer. Precision
data generated for Pb, Sb, Zn, and other metals. Listed detection limits
are based on optimised single-element conditions.
70
-------
20. Chattopadhyay, A. and Katz, S. A., "Determination of ?? Elements In
Geological 5anples by Instrumental Neutron Activation Analysis,"
Radloanal. Chem., 46, pp. 321-331, (1978). INAA determination of up to
22 elements in geological samples. Detection limit data for Cr and Ba and
other metals. Precision within I53t and accuracy for USGS Standards within
10%.
21. Clayton, C. 6. and Packer, 7. W., "Some Applications of F.nergy Dispersive
X-Ray Fluorescence Analysis in Minerals Exploration, Mining, and Process
Control, Advances in X-Pay Analysis, Vol. 21, pp. l-i3, (1978). Discus-
sion of the radiation sources and detectors available for EDXRF, methods
of excitation energy optimization and multielemental analysis applications
to stroam sediments and soils. For multielemental analysis, detection
limits are a function of the excitation energy. Limits are lowered as the
excitation energy approaches the analyte's absorption edge.
22. Chow, T. J., Earl, J. L., Reed, J. H., et al., "Barium Content of Marine
Sediments Near Drilling Sites: A Potential Pollutant Indicator," Har.
Poll. Bull., 9, pp. 97-99, (9178). Concentration levels of 8a 1n
California Sight sediments determined by IDMS and INAA analysis of sample
splits. Correlation between methodologies was 0.923.
23. Church, S. E., "The Importance of the Measurement and Application of
Interelement Corrections to the Analysis of Rocks," Applied Research
Laboratories Publication, pp. 1-24 (September 10, 197571 Evaluation of
1 ntereJemerttai corrections for rnultfelemental analysis of USGS standard
rocks. Author states that correction factors are linear up to 1000 ppm of
the interferent and presents a tabTe of enplHcal spectral interference
values. Accuracy data presented for Be, Ba, Cr, Ni, Cu, Zn and other
metals 1n standard rocks utilizing these empirical correction factors.
24. Cooper, J. A., Wheeler, B. 0., Wolfe, S. J., Bartell, D. M., and SchlafVe,
0. B., "Determination of Sulfur, Ash, and Trace Element Content of Coal,
Coke, and Fly Ash Using Multielement Tube-Excited X-Ray Fluorescence
Analysis," Advances 1n X-Ray Analysis, Vol. 20, pp. 431-436, (1977).
Multielemental determinations of major and trace element concentrations
in pressed pelH-t samples of coal, coke and fly ash. Matrix absorption
corrections were made using interaction coefficients arid analyses of n+2
standards wtere n is the number of Interfering elements.
2£. Currie, L. A., "Limits for Qualitative detection and Quantitative
Determination, Application to Radiochemistry," Anal. Cftero. Vol. 40, pp.
586-593 (1968). Statistical definition of dete'ctlon TTmTts for normal and
Poisson distributions. ICP follows a normal distribution resulting in a
definition of 3.3 x the standard deviation of the responses for the blank.
This "presents a qualitative limit only. Techniques based on counting,
such as XRF and INAA, follow a Poisson distribution (unless the number of
counts 1s large enough to approximate a normal distribution). The detec-
tion limit definition based on a Poisson distribution is slightly dif-
ferent frm the narmal-distritiution definition,
71
-------
26. Davidson, J. W. and Punnell, R. B., "Technical Application Note No. 1,
Multielement Analysis of Marine and Freshwater Sediments by Inductively
Coupled Argon Plasma Spectrometry," J^rrell-Ash Plasma Newsletter, Vol. 3,
No. 2, pp. 4-5 (April 1930). Determination of ba, Cd, Cr, Cu, Ni, Pb, Zn
and other trace and major elements i»; certified soil utilizing a HCI/HNO3
leaching procedure. Spectral interferences were corrected with inter-
elemental correction factors generated empirically.
27. Denmali, M., Ravnikv V., and Kosta, L., "A Fast Iso^jtion and Determina-
tion of Cd in Some Fungi, other Biological Materials, Soil, and Zn Metal
by NAA," Radiochem. Radioanal. Letters, 24 (2), pp. 91-102, (1976).
Analysis oF biological standard"reference materials, soil, and Zn metal
using NAA.
28. Hdiger, R. D-, and Fernandez, F. J., "Optimization of ICP Background
Correction Parameters," Atomic Spectroscopy, Vol. 1, No. 1, pp. 1-7
(January-February 1980). Discussion of spectral interferences due to line
coincidence, line wing overlap, spectral continuum and spectrometer stray
light. Suggested solutions are selection of an alternate line or use of a
background correction system. Precision and accuracy data are given for
As in Fuel and Oil Shale process water based on an ASTM round-robin study.
29. Ediger, R. D. and Wilson, D. L., "The Performance of an Inductively
Coupled Plasma on the Model 5000 Atomic Absorption Spectrophotometer,"
Atomic Absorption Newsletter, Vol. 18, No. 2, pp. 41-45 (March-April
1979). The Perkin-Elmer 5000 ICP is compared to conventional flame and
furnace AA in terms of precision, interferences, sample throughput and
cost. Comparative detection limits given for Ba, Cd and other metals.
Discussion of specific ICP interferences and background correction
procedures.
30. Ellis, K. M. and ^hattopadhyay, A., "Multielement Determination in
Estuarlne Suspended Particle Matter by Instrumental Neutron Activation
Analysis," Anal. Chen., 51 (7), pp. 942-947, (1979). Determination of up
to 28 elements in estuarlne suspended particulate matter. Average pre-
cision varies between 10-16% and the accuracy is witnin 10% of certified
values. Detection limits ranged from 0.37 ng-1.0 ug. Reports spectral
interference of 24^a on As.
31. Fabbi, B. F., "X-Ray Fluorescence Determination of Barium and Strontium
in Geologic Samples," Applied Spectroscopy, Vol, 25, No. 3, pp. 316-318,
(1971). Determination of Ba and Sr in standard reference rocks and min-
erals. Ti interferes with the Ba LI line and a linear correction factor
is applied. Sr analysis must be corrected for Fe absorotion. Detection
limits are given at the 95% confidence level along with accuracy data for
the reference rock and mineral analyses.
72
-------
32. Fassel, V. A., "Quantitative Elemental Analyses by Plasma Emission Spec-
troscopy," Science, Vol. 202, pp. 183-191 (October 13, 1978). Description
of the mechanics of ICP-plasma formation, sample preparation and sample
injection. Discussion of ICP's linear dynamic range and relative freedom
from matrix interferences. Detection limits (3 times the background)
given for all priority pollutant metals ex.cept Ag. Comparison of 1 CP
with other raultielemental methods.
33. Fassel, V. A., "Current and Potential Applications of Inductively Coupled
Plasma (ICP) - Atomic Emission Spectroscopy (AES) in the Exploration,
Mining, and Processing of Materials. Pure and Appl. Chem., Vol. 49, pp.
1533-7.545 (1977). Review article on ICP containing description of
p'asmas, sample prep techniques and discussion of interelement interfer-
ences. References recent publications on analytical applications of ICP.
34. Fassel, V. A., "Inductively Coupled Plasma - Atomic Emission Spectroscopy:
An Alternative Approach to "Flameless" Atomic Absorption Spectroscopy," in
Flameless Atomic Absorption Analysis: An Update. American Society for
Testing and Materials Special Technical Publication 618, 1975, pp. 22-42
(1977). Description of plasma generation and properties. Nebulizer
modifications which enable analysis ot microliter samples. Comparison of
detection limits achieved with ICP and non-flame techniques for As, Be,
Cd, Cr, Cu, N1, Pb, Zn plus other metals.
35. Fassel, V. A. (Chairman), "IUPAC Commission on Spectrochemical and Other
Optical Procedures for Analysis. Nomenclature, Symbols, Units, and Their
Usage in Spectrochemical Analysis - II. Data Interpretation," Anal.
Chen., Vol. 48, No. 14, pp. 2294-2296 (1976). IUPAC definitions given for
statistical nomenclature applied to spectrochemical analyses. Detection
limit at a 90% confidence Interval 1s defined as 3 x the standard deviation
of the background.
36. Fassel, V. A. and Kniseley, R. N., "Inductively Coupled Plasma - Optical
Emission Spectroscopy," Anal. Chem., Vol. 46, No. 13, pp. 1110A-1120A and
1155A-1164A (November 1974). Compilation of detection limits observed for
ICP and flame AAS for all priority pollutant metals plus others. Varia-
tions 1n detection limit definitions are noted. Discussion of interele-
ment effects and the formation and stabilization of plasmas.
37. Feather, C. E., and Willis, J. P., "A Simple Method for Background and
Matrix Correction of Spectral Peaks 1n Trace Element Determination by X-
Ray Fluorescence Spectrometry," X-Ray Spectro»netry, Vol. b, pp. 41-48,
(1975). Presentation of methodology for rapid and accurate background
correction. Since any chosen background 1s Inversely related to the mass
absorption coefficient, background beneath several spectral peaks can be
determined by measurement of only one interference-free background. Rb,
Sr, Th, Ga and Pb in reference rocks are determined using different back-
ground correction methods.
73
-------
38. Feely, R. A. and Massoth, G. J., "Major and Trace Element Composition of
Suspended Matter 1n the North-East Gulf of Alaska: Relationships with
Major Sources," Marine Chemistry, Vol. 10, pp. 431-453, (1981). A study
of the variations in chemical composition and distribution of suspended
particulate matter 1n the northeast Gulf of Alaska. Precision and
accuracy data reported for Cr, Ni, Cu, Zn and major elements in standard
reference rock determined by XRF.
39. Florkowskl, I., Kuc, T., and Piorek, S., "Determination of Trace Elements
1n Plants by X-Ray Fluorescence Analysis for Environmental Pollution
Investigations," In. Jour, of Applied Radiation and Isotopes, Vol. 28, pp.
679-686, (1977). Plant material (prepared either as dried and pressed
pellets or ashed and deposited on filters) were analyzed by XRF for
pollutants. Analysis of NBS Orchard Leaves (performed as a check on the
calibration procedure) provided precision and accuracy data for Cu, Zn,
As, Pb and other metals.
40. Floyd, M. A., Fassel, V. A., and D'Silva, A. P., "F.nvironmental Samples by
Inductively Coupled Plasma - Atomic Emission Spectrometry," Anal. Chem.,
Vol. 52, No. 13, pp. 2168-2173 (November 1980). Determination of several
of the EPA-regulated metals in gecchemlcal and environmental matrices
utilizing a NaOH/HCl fusion-dissolution preparation and a scanning-mono-
chromator Instrument.
41. Franzinl, M., Leonl, I., and Saitta, M., "Determination of the X-Ray Mass
Absorption Coefficient by Measurement of the Intensity of Agk Compton
Scattered Radiation," X-Ray Spectrometry, Vol. 5 pp. 84-87, (1976).
Determination of the relationship between the Intensity of AgK Compton
scattered radiation and the mass absorption coefficient of rock and mineral
samples.
42. Franzinl, M., Leonl, L., and Saitta, M., "A Simple Method to Evaluate the
Matrix Effects in X-Ray Fluorescence Analysis," X-Ray Spectrometry, Vol.
1, pp. 151-154, (1972). Determination of trace metals In USGS reference
silicate rocks. Matrix absorption corrections were made by calculating
absorption coefficients of major elements from artificial standards and
then applying these coefficients to the rock samples once the major
element composition had been determined.
43. Gallor1n1, M. and Orvlni, E., "Determination of Z1nc 1n Environmental
Matrices: A Comparison of Results Obtained by Independent Methods," NBS
Special Publication 422, Accuracy in Trace Analysis: Sampling, Sample
Handling, and Analysis. Preceedings of the 7th IMR Symposium, Galthers-
burg, MD, October 7-11 (1974). Evaluation of INAA, XRF and AAS for the
determination of Zn in NBS Coal Fly Ash and an environmental sample of
Urban Particulates. Precision and accuracy data generated for each method
of analysis. Low AAS values are possibly due to analytp loss during
dissolution.
74
-------
44. Gallorinl, M. and Orvlni, E., "Determination of Zinc in Environmental
Matrices: A Comparison of Results Obtained by Independent Methods,"
National Bureau of Standards Special Publication 422, Accuracy in Trace
Analysis: Sampling, Sample Handling, and Analysis, Proceedings of theTth
1 MR Symposium, Held October 7-11, 1974, Gaitnersburq, MD, pp. 239-245.
(1576).
45. Glauque, R. D., Garrett, R. B., and Goda, L. Y., "Energy Dispersive X-Ray
Fluorescence Spectrometry for Determination of Twenty-Six Trace and Two
Major Elements in Geochemlcal Specimens," Anal. Chan., Vol. 49, pp. 62-67
(1977). Pressed pellets of geochemical reference materials were analyzed
using Compton scattered excitation radiation as an internal standard and
as a measure of spectral background intensity. Precision and accuracy
data presented for most of the priority pollutants plus other elements by
XRF and INAA.
46. Glauque, R. D., Garrett, R. B., and Goda, L. Y., "Determination of Forty
Elements in Geochemic?l Samples and Coal Fly Ash by k-Ray Fluorescence
Spectrometry," Anal. Chem., Vol. 49, pp. 1012-1017, (1977). Presentation
of precision and accuracy data for 40 elements in geochemical samples and
coal fly ash. Matrix absorption effects are minimized by mixing sulfur
powder with the pulverized sample prior to formation of pressed disks.
47. Glauque, R. D., Garrett, R. B., and Goda, L. Y., "Calibration of Energy
Dispersive X-Ray Spectrometers for Analysis of Thin Environmental
Samples," in X-Ray Fluorescence Analysis of Environmental Samples,
Thomas G. Dzubay, Ed., Ann Arbor Science Publishers, Michigan, 1977, pp.
153-164. Description of several calibration techniques available to
standardize for analyses of thin environmental samples.
48. Glaser, J. A., Foerst, D. I., McKee, G. D., Quave, S. A., Budde, W. L.,
"Trace Analyses for Wastewaters," E.S.&T., Vol. 15, No. 12, pp. 1426-1435
(1981). Discusses definition of detection limit based on hypothesis
testing and selection of a specific confidence level. Also provides
analytical procedure for evaluating detection limits independent of
instrumentation or methodology.
49. Gluskoter, H. J., Ruch, R. R., Miller, W. G., et al., "Trace Elements In
Coal: Occurrence and Distribution," 111.State Geol. Survey Circular 499,
pp. 1-154, (1977). Analysis of coal by INAA, AA, XRF and other tech-
niques. Average RSD for any technique 1n range of 10-20% for most elements.
Matrix Interferences for some elements using XRF. INAA resolution pro-
blems for Zn and high bias for Cr. XRF gave high bias for Cu and Ni.
75
-------
50. Gordon, G. E., Randle, K., Goles, G. G., et al., "Instrumental Activa-
tion Analysis of Standard Rocks with High-Resolution Gamma-Ray Detectors,"
Geochim. c-1: Cosmochim. Acta., 32, ppl 369-396, (1968). Investigation of
INAA capabliities for igneous rock with Ge(Li) detectors. List of major
interferents and precision and accuracy for Ba, Cr, Sb and others.
Detection limits for short-lived nuclides dependent on Mn and Na content.
Limits for longer-lived nuclides dependent on Co, Fe and Sc contents.
Suggests use of large and small Ge(Li) detectors or Na1(T1)-Ge(Li) coinci-
dence counting to improve detection limits.
51. Horner, B., "Inductively Coupled Plasma (ICP) in a Precious Metal
Laboratory" in Proc. Int. Precious Metal Institute Conf., pp. 125-130
(1981). Evaluation of ICP as < means to Increase productivity but
maintain comparability with AAS-generated data in a precious metals lab.
Reproducibility data given for Cu aqueous solution. Detection limits
determined for Ag, As, Ba, Be, Cd, Cr, Cu, Hg, Ni, Sb and Zn. Inter-
element interference correction given for Ti effect on Ag.
52. Hutton, J. T. and Elliott, S. M., "An Accurate XRF Method for the Ana-
lysis of Geochanical Exploration Samples for Major ai;d Trace Elements
Using One Glass Disc," Chan. Geo!., Vol. 29, pp. 1-11, (1980). Major and
trace element analysis of geochemical samples using lithium borate fusion
and one glass disc. Detection limits based on 2 x background are given
for Cu, Zn, Ba, Pb and other elements. Precision and accuracy data are
generated and compared with the results obtained by another worker using
pressed pellets of ground rock with binder.
I
53. Irons, R. D., Schenk, E. A., and Giauque, R. D., "Energy Dispersive
X-Ray Fluorescence Spectroscopy and Inductively Coupled Plasma Emission
Spectrometry Evaluated for Multielement Analysis in Complex Biological
Matrices," Clin. Chem., Vol. 22, pp. 2018-2024, (1976). Energy-dispersive
XRF and ICP were evaluated as routine methods for multielemental analysis
of biological matrices. Comparison was based on precision, accuracy, and
detection limits obtained for 13 elements by each technique.
54. Irons, R. D., Schenk, E. A., and Glauque, R. D., "Energy-Dlspersive X-Ray
Fluorescence Spectroscopy and Inductively Coupled Plasma Emission
Spectrometry Evaluated for Multielement Analysis in Complex Biological
Matrices," Clinical Chemistry, Vol. 22, No» 12, pp. 2018-2024 (1976).
Evaluation of energy dispersive XRF and ICP analysis of NBS Bovine Liver
and Orchard Leaves for Zn, Pb, N1, Cu, As, Se and other elements. For ICP
a 1:1 HNO3/HCIO4 digestion was used. Compilation of detection limits,
precision and accuracy. For both techniques, precision found to be
limited by operator error.
55. Jones, J. B., Jr., "Elemental Analyses of Biological Substances by Direct-
Reading Emission Spectroscopy," Jarrell-Ash Plasma Newsletter, Vol. ], No.
1, pp. 4-8 (January 1978). Precision and accuracy data for determination
of Pb, Zn and other metals in NBS Orchard Leaves Involving HNO3 leaching
of ashed material.
76
-------
56. King, B. S., Espos, L. F., and Fabbl, B. P., "X-Ray Fluorescence for
Minor and Trace-Element Analyses of Silicate Rocks In the Presence of
Large Interelement Effects," Advances In X-Ray Analysis, Vol. 21, pp.
75-88, (1978). Precision, accuracy and detection 1 imit data for 16 minor
and trace elements in geological materials. A modified sample-preparation
method was developed to extend pellet life. Multiple linear-regression
equations were used to correct for matrix absorption and interferences.
57. Knechtel, R., Conn, K., annd Fraser, J., "The Analysis of Chemical
Digester Sludges for Metals by Several Laboratory Groups, "Environmental
Protection Service, Fisheries and Environmental, Canada Report No. EPA
4-WP-78-1, pp. 1-38 (January 1978). Interlaboratory comparison of four
sewage treatment plant sludges using AAS with various sample preps, XRF,
INAA and flame emission spectroscopy. Instrumental methods could not be
statistically compared because of insufficient data. Individual data for
Zn, Cu, N1, Cd, Pb, Cr and Hg by INAA and/or XRF can be compared with
mean of AAS deterroinations by the reader. Improvement of intralab pre-
cision over Interlab precision Indicative of systematic differences among
techniques.
58. Kulmatov, R. A. and K1?*, A. A., "Physical Methods of Investigation,
Multielement Instrument Neutron - Actuation Analysis of Water Bodies and
Its Applications in Ecological Investigation," Industrial Laboratory,
44(12), pp. 1482-1485 (December 1978). Determination of toxic metals 1n
water samples. Detection limits based on 3 x background. Precision based
on 7-9 analyses of same sample was found to range from 10-25%. Instru-
mental throughput rate = 20-80 samples/day.
59. Larson, G. F., Fassel, V. A., Scott, R. H., and Knlseley, R. N., "Induc-
tively Coupled Plasma - Optical Emission Analytical Spectrometry. A Study
of Some Interelement Effects," Anal. Chem., Vol 47, No. 2, pp. 238-243
(February 1975). Study of 1nterelement 1nterferences occurring with ICP.
Focuses on CaPOi and Ca-AI Interference system as well as Interference
effects produced by ea$1ly-1on1zed elements.
60. Laul, J. C., "Neutron Activation Analysis of Geological Materials," Atomic
Energy Review, 173, pp. 603-695, (1979). Defines factors affecting
sensitivity (efficiency of detector, irradiation time and neutron flux).
Gives relative detection limits for 60 elements. Discussion of detectors,
precision, accuracy, Interferences and sample prep. Presents accuracy
data for USGS and IAEA soil samples.
61. Lechler, P. J. and Le1n1nger, R. K., "Technical Applications Note No. 2,
Analysis of Black Shale by ICAP Spectroscopy," Jarrell-Ash Plasma News-
letter, Vol. 2, No. 1, ppr 8-10 (January 1979). Presents accuracy data
for some of the EPA-regulatea metals 1n black shale samples based on
comparison with data derived from other labs and AAS. Digestions are
performed 1n unsealed Teflon bombs preventing analysis of volatile^ (As
and Se). Cd and Pb required interelement correction factors.
77
-------
62. Leoni, L. and Saitta, M., "Matrix Effect Correlations by Agk Compton
Scattered Radiation in the Analysis of Rock Samples for Trace Elements,"
X-Ray Spectrometry, Vol. 6, pp. 181-186, (1977) Presentation of accuracy
data for Ba, Cr, Ni, Pb and other metals in standard reference rocks.
Matrix absorption corrections were based on mass absorption coefficients
determined by measuring the intensity of the AgK Compton scattered
radiation.
63. Maenhaut, W. and DeReu, L., "Trace Element Analysis of Biological
Materials: Complementarity of PIXE and NAA, "IEEE Trans, on Nuclear
Science, NS-28(2), pp. 1386-1391, (1981). Precision data for Se an? Zn
analysis of human serum. Large Na and CI concentrations obliterated all
short-lived radionuclides. Precision data for Sb analysis of lung tumors
end Ba and Cr analysis of human albunin.
64. Malyshev, V. I., Shiryaeva, Z. A., Sokolova, L. M., et al., "Application
of Neutron Activation Aralysis to determine the Contents and Isotopic
Ratios of Elements in Rocks and Minerals," J. Radioanal. Chem., 57(2), pp.
287-297, (1980). Determination of 7-14 elements in rock and^. ;ral
matrices. Greatest interferences came from ^Fe, 59pe> an(j fission
fragments. Present precision and accuracy data for As and Sb.
65. Maney, J. P., Luciano, V., and Ward, A. F,, "Technical Application Note
No. 3, Analysis of Energy Resources by ICAP Spectroscopy, Part III:
Uranium," Jdrrell-Ash Plasma Newsletter, Vol. 2, No. 1, pp. 11-13 (January
1979). Tabulation of detection limits for Ni, Pb, Zn and other metals in
water and .n the presence of 1000 ppm U reveals significant deterioration
in detection when U is present. A U extraction procedure is devised which
lowers the detection limits.
66. McGinley, A. N. and Schweikert, E. A., 'Neutron Activation Analysis of
Flints from the Edwards Formation," J. Radioanal. Chem., 52, pp. 101-110,
(1979). Determination of As 1n flint and standard rocks. As not detected
in USGS rocks because the 14-hour irradiation created high matrix activity
which masked the peak. Conclude that USGS reference rocks do not resemble
flints closely enough to allow the Irradiation and counting sequence
chosen for the flint sample to be adequate for the USGS rock standards.
67. McKinney, G. L. and Schlicht, G. R.. "Technical Application Note No. 1,
Analysis of Trace Metals in Fish Tissue by Inductively Coupled Argon
Plasma Emission Spectrometry," Jarrell-Ash Plasma Newsletter, Vol. 3, No.
4, pp. 4-6 (October 1980). Data on spiked recoveries for As, Ba, Cd, Cr,
Cu, Ni, Pb, T1, Zn and other metals in fish tissue digested with nitric
acid. All recoveries were >90% with the exception of T1 (63%). Se was
apparently lost during digestion.
78
-------
68. McKnown, D. M. and Morris, J. S., "Seleniimi Analysis Methodology and
Applications," 11th Conf. on Trace Substances in Environmental Healtl..
Columbia, MO, June 7-9, 1977, pp. 338-344. Discussion of INAA as a single
element technique for analysis of low level Se in biological samples. Use
of "mSe product instead of the more commonly used ''Se avoids the need
for employing radiochemical separation prior to analysis. Precision and
accuracy data given for biological matrices.
69. McLaren, J. W., Beraan, S. S., Boyko, V. J., and Russell, D. S., "Simul-
taneous Determination of Major, Minor, and Trace Elements in Marine
Sediments by Inducively Coupled Plasma Atomic Emission Spectrometry,"
Anal. Chem., 53, pp. 1802-1806 (1981). Presentation of precision and
accuracy data for trace metals in marine sediment reference materials
digested with HNO3/HCIO4/HF in teflon bombs. Cr results were low due to
incomplete dissolution. Cd concentrations occuring in the sediments
were below the detection limit. Other metal results were good indicating
that the use of a simple one or two-point background correction and use of
linear correction factors for spectral interferences were adequate.
70. McQuaker, N. R., Brown, D. F., and Kluckner, P. D., "Digestion of
Environmental Materials for Analysis by Inductively Coupled Plasma -
Atomic Emission Spectrometry," Anal. Chem.,, Voi. 51, No. 7, pp. 1082-1084
(June 1979). Variation in sample salt content or residual digestion acid
can cause substantial analytical errors with ICP. This article investi-
gates digestion procedures which are appropriate for ICP in terms of
minimizing this error and also achieving adequate extraction efficiencies.
Precision and accuracy data are presented for NBS Orchard Leaves, NBS
Bovine Liver, and reference soil material using an HNO3 pre-digestion with
subsequent HF and HCIO4 digestion which provides for a residual acid
content of 3.5 ± 0.4% HCIO4.
71. McQuaker, N. R., Kluckner, P. D., and Chang, G. N., "Calibration of an
Inductively Coupled Plasma - Atomic Emission Spectrometer for the Analysis
of Environmental Materials," Anal. Chem., Vol. 51, No. 7, pp. 888-895
(June 1979). Devise a suitable calibration scheme for ICP variable acid
concentration effects, interelement interference corrections, instrumental
drift, selection of an adequate calibration curve and preparation of
suitable multielemental standards. Calibration performance was evaluated
based on precision and accuracy obtained from analysis of reference
material.
72. Motooka, J. M., Mosier, E. L., Sutley, S. J., and Viets, J. 6., "Induction-
Coupled Plasma Determination of Ag, Au, Bi, Cd, Cu, Pb and Zn in Geologic
Materials Using the Selective Extraction Technique - Preliminary
Investigation," Applied Spectroscopy, Vol. 33, No. 5, pp. 456-460 (1979).
Use of an organic extraction technique to minimize major element Inter-
ferences in analysis of geologic materials for Ag, Au, Bi, Cd, Cu, Pb and
Zn. Use of this organic solvent raises detection limits for Cd, Pa, and
Zn relative to an aqueous solvent. Precision and accuracy relative to AAS
were determined for 6 samples. Results were obtained without any attempt
to correct for background for spoctral interferences yielding accuracies
within 15% of the true value.
79
-------
73. Mudra, K. and Slmkova, M., "Instrumental Neutron Activation Analysis of
Fluorite by Means of Short Activation," Radiochem. Radioanal. Letters,
39(3), pp. 181-188 (1979). Determination of major and minor, components of
fluorite by means of short activation. This material is expected to
become a geological standard after sufficient analysis. The only EPA-
regulated metal present was Ba. Presentation of precision data and detec-
tion limits based on the definition of Currie.25
74. Nielson, K. K., "Matrix Corrections for Energy Dispersive X-Ray Fluores-
cence Analysis of Environmental Samples with Coherent/Incoherent Scattered
X-Rays," Anal. Chem., Vol. 49, pp. 641-648, (1977). Presentation of a
numerical method for calculating matrix absorption in thick pelletized
samples which accounts for light-element as well as heavy-element absorp-
tion. Precision and accuracy data for As, Cu, Ni, Pb, Zn and other
elements in NBS Orchard Leaves and Bovine Liver are derived using this
matrix absorption correction method.
75. Norish, K. and Chappell, B. W., "X-Ray Fluorescence Spectrography in
Physical Methods in Determination Minerology, Ed. J. Zussman, Academic
Press, London (1967), pp. 161-214. A chapter covering x-ray production,
XRF equipment components, qualitative analysis, theory and methods of
quantitative analysis, statistics of x-ray measurement and a discussion
of sample preparation.
76. Norrish, K. and Hutton, J. T., "An Accurate X-Ray Spectrographic Method
for the Analysis of a Wide Range of Geological Samples," Geochim. et
Cosmochim. Acta. Vol. 33, pp. 431-453, (1959). Determination of major
elements in geological samples prepared by fusion with lithium borate
containing lanthanum oxide. Precision and accuracy data for standard
rocks indicate the major source of error is due to counting errors.
77. Odegard, M., "The Use of Inductively Coupled Argon Plasma (ICAP) Atomic
Emission Spectroscopy in the Analysis of Stream Sediments,"Journal of
Geochemial Exploration, 14, pp. 119-130 (1981). Analysis of nitric acid
extracts of stream sediment samples for major and minor elements. Use of
Li or Y as internal standards considerably reduced relative standard
deviations. Detection limits are based on I x standard deviation of tha
blank. Accuracy «»nd precision data presented for synthetic test solutions
and accuracy data given for stream sediment samples.
78. Ohls, K. and Sommar, D., "Analytical Applications of an Air/Argon ICP
Source in Emission Spectroscopy," ICP Information Newsletter, Vol. 4, No.
12, pp. 532-536 (May 1979). Studied substitution of air for the nitro-
gen coolant gas of a nitrogen/argon ICP discharge. Detection limits
lowered for Be, Cu, NI, Zn and several major elements but increased for
B, Ca, Mr and Pb.
80
-------
79. Olson, K. W., Hass, W. J., Jr., and Fassel, V. A., "Multielement Detection
Limits and Sample Nebulization Efficiencies of an Improved Ultrasonic
Nebulizer and a Conventional Pneumatic Nebulizer in Inductively Coupled
Plasma - Atomic Emission Spectrometry," Anal. Chem. Vol. 49, No. 4, pp.
632-637 (April 1977). Comoarison of detection limits obtained with an
improved ultrasonic nebulizer and a pneumatic nebulizer for 14 elements
present in an acidic aqueous matrix. Ultrasonic nebulization with
desolvation gave detection limits superior by an order of magnitude to
those obtained with pneumatic nebulization. The ultrasonic nebulization
performance remained superior even in the presence of NaCl.
80. Pel la, P. A., Lorber,, K. E., and Hc-inrich, K. F. J., "Energy-Dispersive
X-Ray Spectrometric Analysis of Environmental Samples after Boarate
Fusion," Anal. Chem., Vol. 50, pp. 1268-1271, (1978). Application of an
automated borate fusion procedure to N8S Fly Ash and NBS Particulate
Matter. The Fly Ash was fused with heavy absorbers, and the analysis
was conducted assuming no matrix effects. The Particulate Matter was
fused without heavy absorbers and the data were corrected for absorption
and interferences by an iteration method. Detection limits and precision
and accuracy presented for Cr, Cu, Ni, Pb, Zn and other metals.
81. Pongar, R. W. and Tompson, G. R., "Inductively-Coupled Argon Plasma
Emission and Atomic Absorption Spectroscopy: A Comparison Study of
Process Water Sample Analysis," Jarrell-Ash Plasma Newsletter, Vol. 1, No.
22, pp. 5-8 (April 1978). AAS and ICP performance were compared based on
major-element analysis of process water samples. Included in the paper is
a table indicating detection limits and maximum analyzable concentration
for trace and major elements by ICP and AAS. A table of ICP interelement
correction factors is also presented.
82. Perlman, K. and Asaro, F., "Pottery Analysis by Neutron Activation,"
Archaeometry, 11, pp. 21-52, (1969). Analysis of archeologlcal pottery
using a spited, pre-analyzed pottery standard to maximize precision and
accuracy. Enumeration of sources of error and definition of interfering
peaks and background for several EPA regulated metals in this matrix.
Precision data and comparison of INAA results with other techniques.
83. Reynolds, R. C., "Matrix Corrections in Trace Element Analysis by X-Ray
Fluorescence: Estimation of the Mass Absorption Coefficient by Compton
Scattering," Aroeri can Mineraloqist, Vol. 48, pp. 1133-1143, (1963).
Determination of precise mass absorption coefficients from the Compton
scattered portion of a Mo Ka primary beam. Applying this procedure, a
single standard material can serve as a reference standard for samples
with a similar matrix.
84. Robinson, A. L., "Elemental Analysis: Plasmas Revive Emission Spectros-
copy," Science, Vol. 199, pp. 1324-1326 (March 24, 1978). Review of
current state-of-the-art instrumentation with empnasis on cost benefits cf
ICP and particular applications capabilities.
81
-------
85. Rutledge, B. E. and McClurg, J. E., "Technical Aid Note No. 1, Plant
Tissue Analysis by Inductively Coupled Argon Plasma Spectrometry,"
Jarrell-Ash Plasma Newsletter, Vol. 3, No. 3, pp. 4-5 (July 1980). Pre-
sentation of precision and accuracy data for Cu, Zn plus major elements
present in NBS orchard leaves digested with HC1 and HNO3.
86. Schaohter, M. M. and Boyer, K. W., "Digestion of Organic Matrices with a
Single Acid for Trace Element Determination," Anal. Chem., Vol. 52, No. 2,
pp. 360-364 (1980). Review of currently avail aFTe wet and dry ashing
techniques for destruction of organic matrices prior to tr?ce element
analyses. Authors recommend a nitric acid wet digestion using a specified
digestion apparatus. The method was tested by discerning recoveries and
precision obtained for 6 biological NBS reference materials. Low re-
coveries wera obtained for Cr in Orchard Leaves and Spinach and high
recoveries were obtained for Pb in Bovine Liver and Spinach. The pre-
sence of chlcride or high concentrations of Ca, Mg, and P did not seri-
ously affect recoveries.
87. Scheer, J., Voet, L., Watjen, U., et al., "Comparison of Sensitivities in
Trace Element Analysis Obtained by X-Ray Excited X-Ray Fluorescence and
Proton Induced X-Ray Emission," Nuclear Instruments and Methods, Vol. 142,
pp. 333-338, (1977). Comparison of trace element detection limits b) XRF
and PIXE. Sensitivity of the XRF system is much more matrix independent,
than the PIXE.
88. Schleicher, R. G., Leighty, D. A., Kahn, H. L. and Sotera, J. J., "The
Analysis of Difficult Samples by Plasma Spectroscopy," Instrumentation
Laboratory, Inc. Report Number 20 (July 1980). Presentation of detection
limits and instrumental precision. Discussion of methods development for
instrument parameter settings. Precision and accural j data for determi-
nation of Ba in an NBS seawater standard were good due to lack of spectral
interferences and very little background emission. Determination of Al in
wood pulp and trace metals In fuel oils also presented.
89. Schroeder, B., Thompson, G., Sulanowska, M., "Analysis of Geologic
Materials Using an Automated X-Ray Fluorescence System," X-Ray Spectrom-
etry , Vol. 9, pp. 198-205, (1980). Analysis of geologic materials using
an automated XRF system. Major elements are determined by fusion while
trace elements are determined directly on the rock powder. Correction for
'Tiatrix absorption is done off-line utilizing previously determined major
element concentrations and published interelement correction factors.
Precision and accuracy data are reported for standard rocks.
90. Scott, R. H. and Kokot, M. L., "Application of Inductively Coupled Plasmas
to the Analysis of Geochemical Samples," Anal_. Chim. Acta., 75 pp. 257-270
(1975). Analysis of geochemical soil samples by ICP and AAS using an
HNO3/HCIO4 digestion for Co, Cu, Ni, Pb and Zn. Report detection limits
based on 3 x background, precision as a function of concentration level,
and a regression plot of ICP versus AAS values.
82
-------
91. Spencer, D. W., Brewer, P. G., and Sachs, P.I., "Aspects of the Distribu-
tion and Trace Element Composition of Suspended Matter in the Black Sea,"
Geochim. et Cosmochim. Acta., 3€, pp. 71-86, (1972). Analyses of sus-
pended matter from the Black Sea for Cr, Hg, Zn, and otner metals but no
precision or accuracy data is presentee.. Detection limits reported as 3
x background for blank Millipore filters.
92. Stross, F. K., Sheets, P. Asaro, F., and Michel,.H. V., "Precise Charac-
terization of Guatemalan Obsidian Sources, and Source Determination of
Artifacts from Quingua," Submitted to American Antiquity, (November
1981). Discussion of requirements for accurate lab mtercalibration.
Labs must use the same standard or well known standards and background and
interference corrections must be uniformly made. Interferences are listed
for Ag, As, Ba, Cu, Cr, Ni, Zn and others. Precision data.for Ba and
other metals in obsidian samples. Accuracy defined as the square root of
the sum of the squares of the precision and the accuracy of that element
in standard pottery.
93. Thomas, I. L. snd Haukka, M. T., "XRF Determination of Trace and Major
Elements Using a Single-Fused Disc," Chem, Geo!., Vol. 21, pp. 39-50,
(1978). Development of a fusion preparation procedure suitable for major
and trace element analysis in a wide range of geological materials.
94. Thompson, M. Imperial College of Science and Technology, London,
England, Personal Communication (July 1981). Detection limit data for
over 30 metals. Thorough discussion of the digestion procedures available
for analysis of geochemical samples and recommendations of suita^e diges-
tions for specific geochemical samples.
95. Thompson, M. and Howarth, R. J., "A New Approach to tlie Estimation of
Analytical Precision," Journal of Geochemical Exploration, 9, pp. 23-30
(1978). Discussion of estimation of analytical precision without assuming
that precision remains independent of concentration level.
96. Ting, 8. T. G. and Mrnahan, S. E., "Effects of Organic and Inorganic
Binding on the Volatilization of Trace Elements during Coal Pyrclysis,"
ES&T, 13, pp. 1S37-154C, (1979). Precision data presented for analysis of
As, Cr, and Se in coal. The remainder of the paper is a study of element
volatility of coal object to pyrolysis.
97. Turekian, K. K., Katz, A., and Chan, L., "Trace Element Trapping in
Pteropod Tests," L&0, 18, pp. 240-9 (1973). Analysis of pteropod and
planktonic samples for trace elements. Detection limits bassd on 2 back-
ground presented for Cr, Sb, Se a"d other elements. Cr precision is due
to counting statistics only.
83
-------
98. Tyler, L. C., "Technical Aiu Note No. 2, Analysis of Sewage Sludge by
Inductively Coupled Argon Plasma Spectrometry," Jarrell-Ash Plasma News-
letter, Vol. 3, No. 3, pp. 6-8 (July 1980). Precision data for six of the
priority pollutants plus other metals in sludge filter cakes. Sample
preparation consisted of dry ashing and ether extractions to remove Fe.
The Fe extraction was necessary becaure at the 5% level the Fe overloaded
the detector and use of an interelement correction factor was no longer
valid.
99. Uchide, H., Uchida, T. and Chuzo, I., "Determination of Minor and Trace
Elements in Silicate Rockes by Inductively-Coupled Plasma Emission
Spectrometry," Anal. Chim. Acta., 116, pp. 433-437 (1980). Presentation
of precision and accuracy data for ICP analysis of major and trace metals
in silicate rocks. The digestion proceduia consisted of HF aqua regia in
a teflon bomb followed by evaporation to dryness. Detection limits were
based on 2 x the standard deviation for the blank.
100. Varnes, A. W. end Andrews, T. E., "Comparison of Wear Metal Analysis of
Used Motor Oil by ICAP Spectroscopy and Rotating Disk Excitation,"
Jarrell-Ash Plasma Newsletter, Vol. 1, No. 1 pp. 12-14 (January 1978).
Comparison of wear metal concentrations in used motor oils analyzed by
ICP.and rotating disk electrode. ICP results were in good agreement
with rotating disk electrode except for A1, Na, Si, Pb, Ni and Sn. ICP
detection limits for wear metals in oil were calculated.
101. Ward, A. F., "Inductively Coupled Argon Plasma Spectroscopy. Development,
Technique, and Applications," American Laboratory (November 1978).
History, performance characteristics, anu instrumentation configuration of
ICP is discussed. Aqueous detection limits are presented relative to
other techniques. Accuracy data presented for liver, rock, steel, and
fuel oil reference materials and samples.
102. Ward, A. F., "Technical Aid Note Nc. 1, Instrument Performance/Diagnosis
Tests. Part I: ICAP Atom Comp Background Correction System." Jarrel1 -
Ash Plasma Newsletter, Vol. 1, No. 4, pp. 14-17 (October 1978). Presenta-
tion of aqueous detection limit data and the effect of Na, K, Mg, Ca, S, N
onti C (at the 10,000-ppm level) on the apparent trace element concen-
trations.
103. Ward, A. r. and Marciello, L., "Technical Application Note No. 2, Analysis
of Energy Resouces by [CAP Spectroscopy, Part II: Petroleum Products,"
Jarrell-Ash Plasma Newsletter, Vol. 1, No. 4, pp. 10-13 (October 1978).
Precision, accuracy and detection limit data are presented for trace
elements in wear oils. IlP values were compared with AAS and XRF for
lubncatinq oils and crude oil. Precision and accuracy data were also
obtained frum analysis of NBS reference fuels.
84
-------
104. Ward, A. F. and Marclello, L., "Technical Application Note No. 2, Analysis
of Energy Resources by ICAP Spectroscopy, Part I: Coal Fly Ash and Coal,"
Jarrell-Ash Plasma Newsletter, Vol. 1, No. 3, pp. 10-13 (July 1978).
Comparison of three different digestion procedures for the analysis of
cool and coal fly ash. Precision and accuracy data generated for 2u
metals In NBS Coal and Coal Fly Ash Indicate that use of a Teflon pressure
bomb and HF gives results which are superior to the results obtained by
bicarbonate fusion or the Teflon-bomb procedure utilizing HCIO4.
105. YelUn, J., Perlman, I., Asaro, F., Michel, H. v., and Mosler, D. F.,
"Neutron Activation Analysis of Complex Mixtures - An Interlaboratory
Study," Unpublished. Test intrinsic reliability of INAA In complex
matrices by performing an Interlaboratory study. Agreement Mas good over
a wide range of composition.
106. YelUn, J., Perlman, J., Asaro, F., Michel, H. V., and Hosier, D. F.,
"Comparison of Neutron Activation Analysis from the Lawrence Berkeley
Laboratory and the Hebrew University," Archaeometry, 20, 1, pp. 95-100,
(1978). Analysis of a series of samples In two laboratories which have
different equipment and modes of operation In order to test for compara-
bility and soutxes of error. Enumeration of the various potential error
sources. Results provided precision data for Ba, Cr, N1 and other
elements.
107. Zamechek, W., "Trace Metals Analysis In SI and A1 Metals by ICAP
Spectroscopy," Jarrell-Ash Plasma Newsletter, Vol. 1, No. 1, pp. 15-20
(January 1978). Determination of trace metals 1n S1 and A1 metals and
their oxides. Detection limits are given In aqueous matrix and as a
function of an Increasing A1 matrix. Percent recoveries are also given as
a functUn of A? concentration. An Increase 1n the A1 matrix has the
expected effect of decreasing recoveries and Increa.ing detection limits.
ICP analytical results on real samples are comparec *0 AAS data.
85
-------
APPENDIX A
INSTRUMENT USER QUESTIONNAIRE
X-Ray Fluorescence - XRF
Instrirantal Neutron Activation Analysis - INAA
Inductively Coupled Plasma - ICP
1. Date Instrument Acquired
Instrument Manufacturer
Model No.
Capital Cost
Expendables cost
(supplies)
Labor Cost
Downtime Estimation _
No. of Staff Members
Involved in this
area of work
_/month
/month
2. What type of sample prep 1s performed prior to the instrument analysis?
3. Interferences
a. Indicate direct Interferences encountei°d 1n the analysis of the
priority pollutant metals (As, Ag, Ba, Be, Cd, Cr, Cu, Hg, Pb, Ni,
Se, Sb, T1, Zn) In environmental matrices. What Is the magnitude of
Interferences (e.g., mg analyte/1000 ppm interferent)?
b. What correction methods do you use for analyzing any of the above
elements when present in a complex matrix containing interferences?
1. Factor (mg Analyte/mg Interferent)?
2. Alternate wavelengths or energies?
3. Other?
4. Comments
86
-------
4. Statistical Section:
a. Detection limit for the following elements (ug/1 or ug/g or ng)
(circle one)
(ICP users please Indicate wavelength)
Ag
at
nm
Hg
at
nm
As
at
nm
N1
at
nm
Ba
at
nm
Pb
at
nm
Be
at
nm
Sb
at
nm
Cd
at
nm
Se
at
Ml
Cr
at
nm
T1
at
nm
Cu
at
nm
Zn
at
nm
b. What 1s your I1»1t of quantitation?
3 x D.L. 10 x D.L. Other
How 1s the D.L. determined? (Please check the appropriate line)
2 x standard deviation of the background noise
3 x standard deviation of the background noise
^ 3 x standard deviation of a low conc. standard
2 x standard deviation ordinate Intercept calculated
from regression of calibration standards
2 x standard deviation background (analyte conc)
signal to background ratio
Other
c. Typical precision observed at
—iO x D.L. Is and 1s based on no. of replicates
100 x D.L. 1s and Is based on no. of replicates
How does dally precision compare with long term preclson (I.e., 1s
Instrument drift a problem?)
d. Are accuracy data obtained from analysis of standard reference
materials or by comparison of different techniques? (List materials
or techniques and elements analyzed)
87
-------
What 1s your rate of sample throughput?
Time limiting step occurs where?
Commercial Labs Only:
Approximate Price Quote/Sample for analysis of these elements:
a) In solids b) In liquids
Comments
Do you wish any of this information to be confidential? (Please star
confidential data) (i.e., only presented in statistical stannaries)
Name of individual to contact for further information:
Lab Name Contact
38
-------
APPENDIX B
LIST Of INDIVIDUALS RECEIVING THE INSTRUMENT USER QUESTIONNAIRE
Commercial Users: Replied:
ICP
Dr. Rau No
Gemini Industries
2311 So. Pullman St.
Santa Ana, CA 92714
Ms. JoAnn Daniel son No
Shasta Analytical Lab
1240 Redwood Blvd.
Redding, CA 96003
Mr. Keith Seace No
Roy F. Weston Inc.
Weston Way
Westchester, PA 19380
Mr. Bob Stromatt No
Westlnghouse Hanford
Chemical and Analytical Dept.
Richland, WA 99352
fts. Carol Jones No
Environmental Services Dept.
Dow Chemical
Freeport, TX 77541
Mr. Bob Adams Yes
American Enka
Research Street
Enka, NC 28728
Ms. Holly Ham No
Data Lab Inc.
2900 Corvln Drive
Santa Clara, CA 95051
89
-------
Commercial Users: Replied:
Mr. Dave Leighty No
Instrumentation Laboratory, Inc.
Jonspln Road
Wilmington, MA 01887
Dean Toye No
Acme Analytical Laboratories
852 East Hastings Street
Vancouver, British Columbia, rjnada
Gordon Utas Yes
Chemical and Geological Labs, Inc.
14203-129 Avenue
Edmonton, Alberta, Canada
Gordon Van Sickle
Skyline Labs
12090 West 50th Place
Wheatridge, CO 80033
No useful information
Derrel Dixon No
Can-Test Limited
1650 Pandora Street
Vancouver, British Columbia, Canada
Mr, John Burgener No
Technical Service Laboratory
516 Unior, Road
P.O. Box 14642
Opportunity, WA 99214
Dr. Debnam No
Technical Service Lao
1301 Fewster Drive
Mississauga, Ontario, Canada L4W 1A2
Atomic Energy of Canada Ltd. - Research No useful information
Whiteshell Nuclear Research Establishment
Pinawa, Manitoba, Canada ROE 1L0
Radian Corp.
8500 Shoal Creek Blvd.
Austin, TX 78753
Sverdrup/ARO, Inc.
Environmental Laboratory
P.O. Box 884
Tullahoma, TN 37388
No
No useful Information
90
-------
Commercial Users:
Orlando Labs
P.O. Box 19127
Orlando, FL 32814
Barringer Resources
1626 Cole Aveneue
Golden, CO 80401
Or. Rick Ronan
Versar Inc.
6621 Electronic Drive
Springfield, VA 22151
XRF
ZGAG Ortec
100 Midland Road
Oak Ridge, TN 37830
Kevex Corporation
1101 Chess Drive
roster City, CA 94404
ptew England Nuclear
549 Albany Street
Boston, MA 02118
Nuclide Corporation
642 E. College Ave.
State College, PA 16801
Philips Electronic Instruments
85 McKee Drive
Mahwah, rw 07430
Tracor Northern, Inc.
2551 W. Beltline Hwy.
Mlddleton, WI 53562
Spectrochem
545 Commerce Street
Franklin Lakes, NJ 07417
American Standards Testing Bureau, Inc.
40 Water Street
New York, NY 1004
Replled:
No
No
No
No
No
No
No
No
No
No
No
91
-------
Commercial Users: Replied:
Applied Research Laboratories of Fla., Inc. No
P.O. Box 489
Hlaleah, Fl. 33011
Coor Spectro-Chemlcal Lab No
D1v. of Coors Porcelain Co.
17750 W. 32nd Ave.
Golden, CO 30401
Gollob Analytical Service No
47 Industrial Road
Berkeley Heights, NJ 07922
National Spectrograph1c Laboratories, Inc. No
7650 Hub Pkwy.
Valley View, OH 44125
NUS Corp. No
4 Research Place
Rockvllle, MO 20850
Princeton Testing Laboratory, Inc. No useful information
US Route 1
Princeton Service Center
Box 3108
Princeton, NO 08854
United States Testing Co. No
1415 Park Ave.
Hoboke.i, NJ 07030
Hazleton No
1500 F'ontage Road
Northbrock, IL 60062
INAA
Atomerglc Chemetals Corp. No
100 Fairchlld Ave.
Plalnview, NY 11803
General Activation Analysis, Inc. Yes
11575 Sorrento Valley Rd.
Bldg. 214
San Diego, CA 92121
92
-------
Commercial Users:
Replled:
Nuclide Corp. No ur.eful information
AGV Div.
P.O. Box 315
Acton, MA 01720
Mr. Robert Strack Yes
ClntlChem, Inc.
P.O. Box 816
Tuxedo, NY 10987
Non-Commercial Users: Replied:
ICP
Mr. R. C. Goldsmith No
Sante Fe Railroad
10001 No. East Atchinson St.
Topeka, KA 66616
Mr. Sam Hopper No
Dept. of Natural Resources
2010 Missouri Blvd.
Jefferson City, MO 65201
Mr. Andy Britol Yes
Agronomy Department
New Mexico State University
Las Cruces, New Mexico 88003
J. W. Jones Yes
U.S. Dept. of Health and Human Services
Public Health Service
Food and Drug Administration
Bureau of Foods
200 C Street S.W.
Washington, DC 20204
D. Taylor
U.S. Dept. of Health and Human Services No
Public Health Service Center for Disease Control
National Institute for Occupational Safety & Health
Division of Respiratory Disease Studies
944 Chestnut Ridge Road
Morgantown, WV 26505
93
-------
Non-Commerciil Users:
P.J. Lamothe
U.S. Dept. of the Interior
Geological Survey, Geologic Division
Branch of Analytical Laboratories
345 Middlefleld Rd.
Menlo Park, CA 94025
J.A. Caruso
University of Cincinnati
McMicken College of Arts & Sciences
Dept. of Chemistry
Van Wormer Adm. Bldg.
Room 204
Cincinnati, OH 45221
F.W. Brown
U.S. Dept. of the Interior
Geological Survey, Geologic Division
12201 Sunrise Valley Drive
Reston, VI 22092
Harvard Taylor
U.S. Geological Survey
5293 Ward Road
Arvada, CO 80002
Dr. Dalvd Erdmann
U.S. Geological Survey
Atlanta Water Quality Lab
6481-H Peachtree Industrial Blvd.
Suite 1
Doravi'ile, GA 30340
Dr. J1m CalUs
University of Washington
Chemistry BG-10
Bagley Hall
Seattle, WA 98195
Dr. P.N. Soltanpowr
Colorado Sate University
Soil Testing Lab
Fort Collins, CO 80523
J1m Seeger
Cit.y of Detroit
Water and Sewage/Wastewater
9300 West Jefferson
Detroit, MI 48209
-------
Non-Commercial Users:
John Bray
East Carolina University
School of Medicine
Department of Surgery
Trace Metals Lab
Greenville, NC 27834
Larry Williams
Monsanto Company
Central Research Building South
800 North Lindbergh Blvd.
St. Louis, MO 63166
Eric Gauglitz
NOAA Marine Fisheries
2725 Montiake Blvd. , East
Seattle, WA 98112
Joan Rummler
Sel-Rex Division
75 River Road
Nutley, NJ 07100
Steve Matza
Shell Development
3737 Bell aire Blvd.
Houston, TX 77025
Dr. Braselton
Michigan State University
1330 South Harrison Road
East Lansing, MI 48824
Jean Guimont
Mlnistre Des Rlchesses Naturelles
Centre de Recherches Minerales
2700 Rue Einstein
Ste-Foy, Quebec, Canada GIP 3W8
Dr. Mark Cherwin
Milwaukee Metro Sewerage District
P.O. Box 2079
Milwaukee, WI 53201
Dr. Dennis Yates
University of Missouri
Environmental Trace Substance Lab
Route 3
Columbia, M0 65201
-------
Non-Comriiercial Users:
Replied:
Dr. Ken Smith
Illinois Natural History Survey
236 NRSA
Urbana, IL 61801
No
Sandy Gallagher
No
U.S. Departn.^nt of Agriculture
Human Nutrition Lab
2420 Second Avenue, N
P.O. Box 7166
University Station
Grand Forks, ND 58201
Richard Leininger No
Indiana State Geological Survey
11th and Walnut Grove
Bloomington, IN 47401
W. H. Patrick, Jr. No
Louisiana State University
Lab for Wetland Soils and Sediments
Baton Rouge, LA 70803
Or. R. B. Swingle No
Environmental Protection Service (FMS)
Department of Fisheries and Environment
Lab Services
4195 Marine Drive
West Vancouver, British Columbia, Canada
Dr. Bob Botto No
Exxon Research and Engineering Company
P.O. Box 4255
Baytown, TX 7752C
Bill Blakimore No
Food and Drug Administration
National Center for Toxicological Research F2620-78
Jefferson, AK 72079
Pat Johnson No
Environmental Protection Agency
Annapolis Field Office
Annapolis Science Center
Riva Road
Annapolis, MD 21401
96
-------
Non-Commercial Users:
Replled:
Dr. H.S. Samant
Yes
Environmental Protection Services
Water Quality Control
Bedfore Institute of Oceanography
Before, Nova Scotia
Canada
Boston College
Chestnut Hill Campus School of Arts & Sciences
Dept. of Physics
140 Commonwealth
Chestnut Hill, UA 02167
Gerry Stummer No
Dept. of Geochemistry
UCLA
Campus Drive
Los Angeles, California
Dr. Neil McQuaker No
Environmental Laboratory
Ministry of the Environment
3650 Wesbrook Crescent
Vancouver, BC V6S 2L2
Canada
J1m Gill No
Marine Science Center
UC Santa Cruz
Santa Cruz, California
Don Hendricks No
University of Arizona
Dept. of Soils
Tucson, AR 85721
XRF
R. H. April
Colgate University
Graduate School
DeDt. of Geology
Hamilton, NY 13346
No useful Information
J. F. Sieckarski
No
97
-------
Non-Commercial Users:
John Halloway
Arizona State
Uept. of Geochemistry
Tempe, AR 85281
Bill Parry
Dept. of Geochemistry
University of Utah
Salt Lake City, Utah
Replled:
No
No useful information
Dr. D1ck Feely Yes
Pacific Marine Env. Lab
7600 Sand Point Way NE
Seattle, WA 98115
Dr. L. S. B1rks No
Naval Research Lab
Washington, DC
Dr. Brian Price No
Dept. of Geology
University of Edinburgh
Edinburgh
Scotland
Mr. Dan Shadoan No
University of California, Davis
Crocker Nuclear lab
Davis, California
INAA
K. F. Flynn No
U.S. Dept. of Energy
Argonne National Lab
9700 S. Cass Ave.
Argonne, IL 60439
J.F. S1eckarsk1 No
Boston College
Chestnut H111 Campus School of Arts & Sciences
Dept. of Physics
140 Commonwealth
Chestnut H111 MA 02167
98
-------
Non-Coanerc1a1 Users:
Dr. A. Chattopadhyay
Trace Analysis Research Centre
Dalhousle University
Halifax, Nova Scotia
Thonas E. Gills
Activation Analysis Section
Analysltcal Chenlstry Division
NBS
Washington, DC 20234
Dr. J. C. Laul
Physical Sciences Dept.
Battelle Northwest
Rlc'ilar-, WA 99532
Dr. Jack Heaver
Nuclear Engineering Lab
North Carolina State University
Raleigh, NC
Dr. J. J. Rove
U.S. Geological Survey
12202 Sunrise Valley Drive
Reston, VA 22092
Dr. E. S. Gladney
University of California
Los Alanos Scientific Lab
KS 490
Los Alanos, NM 87545
Dr. Bill Zoller
Dept. of Chemistry
University of Maryland
College Park, MD
Dr. Derek Spencer
Woods Hole Oceanographlc Institute
Woods Hole, MA 02543
Dr. Robert Duce
Graduate School of Oceanography
University of Rhode Island
Kingston, R1
Replled:
Too late for Inclusion
Yes
No
No
No
No
No
No
No
99
-------
------- |