Hazard Ranking System Issue Analysis:
Laboratory Limits of Detection
MITRE
-------
Hazard Ranking System Issue Analysis:
Laboratory Limits of Detection
Jerry M. Fitzgerald
July 1986
MTR-86W77
SPONSOR:
U.S. Environmental Protection Agency
CONTRACT NO.:
EPA-68-01-7054
The MITRE Corporation
Metrek Division
7525 Colshire Drive
McLean, Virginia 22102-3481
-------
Department Approval:,
MITRE Project Approval:.
-------
ABSTRACT
The U.S. Environmental Protection Agency (EPA) uses the Hazard
Ranking System (HRS) to estimate the effects of the release or
potential release of hazardous substances; this estimate is required
by the Comprehensive Environmental Response, Compensation, and
Liability Act of 1980 (CERCIA). Scoring a site for release or
potential release requires the detection of hazardous substances by
chemical analysis. The Limit of Detection (LOD) for analytical
methods is thus of concern to EPA because a lower value of LOD will
increase the number of different substances reported present at a
site. This technical review addresses two concerns regarding LODs
and their impact on the HRS.
The first concern is: what is the likelihood, in the near
term, that LODs for chemicals which are frequently reported at
National Priorities List (NPL) sites will reach significantly
smaller concentrations? A number of sources were consulted to
arrive at the conclusion that LODs will be reduced only slightly, if
at all, in the near term. Around 1 ppb or so, most methods have so
much inherent imprecision that the number of repetitive analyses
required to measure a given lower concentration is prohibitively
large. Unless there is a special need for data below around 1 ppb,
then an LOD barrier has been reached in essence.
The second concern is: radioactive elements can be measured
either by radiochemical counting or else by methods currently in use
for environmental analysis of metals. There could be a substantial
difference in LOD between the two techniques and hence a method-
dependent bias in reporting the presence or absence of radioactive
elements.
Five radioactive elements, of importance in the HRS program,
were selected for a comparative study. For those radioactive
elements with half-lives of around a billion years, the radio-
chemical LOD values are about the same as those obtained by
spectroscopic methods. But, if the element in question has a
half-life of only a few thousand years, the radiochemical LOD is
around a million times smaller than the spectroscopic LOD. There-
fore, a bias for radiochemical detection exists if radioactive
elements with short half-lives are in a sample from an NPL site.
iii
-------
TABLE OF CONTENTS
Page
LIST OF TABLES vi
1.0 INTRODUCTION 1
1.1 Background 1
1.2. Limits of Detection (LOD) and Estimation
of Hazard 3
1.3 Objectives 4
1.4 Scope and Approach 5
1.5 Organization of the Paper 6
2.0 ORIGIN AND DEFINITIONS OF LOD 9
2.1 Origin of the LOD Concept 9
2.2 Description of the LOD 10
2.3 Calculation of the LOD 11
3.0 REGULATORY SPECIFICATION OF LODs 13
4.0 POTENTIAL FOR IMPROVEMENTS IN LABORATORY LOD VALUES 19
4.1 Past Improvements in the LOD 19
4.2 LOD Limits from Interlaboratory Data 21
4.3 LOD Limits from Proficiency Testing Data 23
5.0 EQUIVALENCE OF LODs FOR RADIOCHEMICAL COUNTING
COMPARED TO SPECTROSCOPIC METHODS 25
5.1 Selection of Elements to be Considered 25
5.2 Quantitative Principles for Radionuclides 26
5.3 Inorganic LODs for Comparison Purposes 27
5.4 LOD Values for Radioactive Elements in
Environmental Samples 27
6.0 CONCLUSIONS 31
6.1 Anticipated Changes in Operational LODs 31
6.2 LOD Comparisons Between Radiochemical and
Spectroscope Techniques 33
REFERENCES 35
-------
LIST OF TABLES
Table Number Page
1 Limits of Detection Required By Two
EPA Programs for Substances Found at
NPL Sites 15
2 Limits of Detection for Radioactive
Substances Found at NPL Sites 30
vi
-------
1.0 INTRODUCTION
1.1 Background
The Comprehensive Environmental Response, Compensation, and
Liability Act of 1980 (CERCIA) (PL 96-510) requires the President to
identify national priorities for remedial action among releases or
threatened releases of hazardous substances. These releases are to
be identified based on criteria promulgated in the National
Contingency Plan (NCP). On July 16, 1982, EPA promulgated the
Hazard Ranking System (HRS) as Appendix A to the NCP (40 CFR 300;
47 FR 31180). The HRS comprises the criteria required under CERCLA
and is used by EPA to estimate the relative potential hazard posed
by releases or threatened releases of hazardous substances:
The HRS is a means for applying uniform technical judgment
regarding the potential hazards presented by a release relative to
other releases. The HRS is used in identifying releases as national
priorities for further investigation and possible remedial action by
assigning numerical values (according to prescribed guidelines) to
factors that characterize the potential of any given release to
cause harm. The values are manipulated mathematically to yield a
single score that is designed to indicate the potential hazard posed
by each release relative to other releases. This score is one of
the criteria used by EPA in determining whether the release should
be placed on the National Priorities List (NPL).
-------
During the original NCP rulemaking process and the subsequent
application of the HRS to specific releases, a number of technical
issues have been raised regarding the HRS. These issues concern the
desire for modifications to the HRS to further improve its
capability to estimate the relative potential hazard of releases.
The issues include:
• Review of other existing ranking systems suitable for
ranking hazardous waste sites for the NFL
• Feasibility of considering ground water flow direction and
distance, as well as defining "aquifer of concern," in
determining potentially affected targets
• Development of a human food chain exposure evaluation
methodology
• Development of a potential for air release factor category
in the HRS air pathway
• Review of the adequacy of the target distance specified in
the air pathway
• Feasibility of considering the accumulation of hazardous
substances in indoor environments
• Feasibility of developing factors to account for
environmental attenuation of hazardous substances in ground
and surface water
• Feasibility of developing a more discriminating toxicity
factor
• Refinement of the definition of "significance" as it relates
to observed releases
• Suitability of the current HRS default value for an unknown
waste quantity
• Feasibility of determining and using hazardous substance
concentration data
-------
• Feasibility of evaluating waste quantity on a hazardous
constituent basis
• Review of the adequacy of the target distance specified in
the surface water pathway
• Development of a sensitive environment evaluation methodology
• Feasibility of revising the containment factors to increase
discrimination among facilities
• Review of the potential for future changes in laboratory
detection limits to affect the types of sites considered for
the NPL
Each technical issue is the subject of one or more separate but
related reports. These reports, although providing background,
analysis, conclusions, and recommendations regarding the technical
issue, will not directly affect the HRS. Rather, these reports will
be used by an EPA working group that will assess and integrate the
results and prepare recommendations to EPA management regarding
future changes to the HRS. Any changes will then be proposed in
Federal notice and comment rulemaking as formal changes to the NCP.
The following section describes the specific issue that is the
subject of this report.
1.2- Limits of Detection (LOP) and Estimation of Hazard
The EPA uses the HRS to estimate the relative potential hazard
posed by releases or threatened releases of hazardous substances.
When the HRS was originally developed, the laboratory limits of
detection (LOD) for substances likely to be found at release sites
were typically in the range of parts-per-billion (ppb) to
-------
parts-per-million (ppm). For purposes of the HRS, EPA considers
hazardous substances detected at these concentrations to pose a
potential for concern. Several of the rating factors used in the
HRS (e.g., observed releases and toxicity) can be affected by the
LOD because these factors rely on a determination that hazardous
substances are identified at sampling locations.
EPA is concerned that if analytical laboratory detection limits
for hazardous substances were to decrease significantly during the
next few years, then substances could be detected at release sites
at levels where they may not pose a potential for concern. If such
a lowering of analytical detection limits was to occur, then it
could be necessary to specify minimum concentrations below which the
hazardous substances could not be used for HRS estimates.
1.3 Objectives
There are two objectives for this report. Both objectives are
united by a concern with the analytical LCDs for chemical substances
frequently found at hazardous sites.
The first objective is to identify the current LOD values for
substances which occur frequently at hazardous waste sites and to
estimate whether the LODs for some or all of these substances will
improve (be lowered) over the next few years.
The second objective is to identify whether substantial
differences might exist in LODs for different methods of quantifying
certain radioactive elements. Large differences in LODs will result
-------
in a bias for detection of some radioactive elements when measured
in different ways. This can affect the HRS score.
1.4 Scope and Approach
Examination of the question of improvements in LCDs for
hazardous substance determination is restricted to those kinds of
procedures currently used in environmental analysis, since it is
these procedures that are commonly used in the site evaluation phase
of the Hazardous Substance Response Plan (40 CFR 300.66). The
presentation begins by tracing the concept of LOD from early use in
the laboratory, and on to more sophisticated scientific treatments.
The interplay of technical ability and regulatory needs are
demonstrated by using LOD criteria from the EPA Contract Laboratory
Program (CLP) and the guidelines under the Clean Water Act.
Finally, the likelihood of major improvements in LODs for the
substances involved in the HRS is discussed. The likelihood of
improvement is based on three considerations.
• How the LODs for current environmental procedures can be
improved
• Compilations of data from interlaboratory analyses which
show an exponential worsening of precision (repeatability of
data) as concentration is decreased
• Proficiency test data for many analyte concentrations which
show degradations of precision as concentrations are lowered
The approach used for all three factors above is to look for
any general trends that could lead to improved LOD values for the
kinds of environmental methods considered in this review.
-------
The consideration of LOD values for radiochemical counting
compared to the more usual analytical chemistry procedures used to
analyze environmental samples for quantification of inorganic
substances will be limited to five elements. These elements are
used as models for the more general case. These five were selected
because they have also been reported at NPL sites. LOD values
obtained from two independent sources are compared against each
other. Comparisons between LODs of different elements is made in
order to identify any trends in sensitivities.
1.5 Organization of the Paper
This report first describes the scientific origins of the LOD
concept and then addresses the formal definition of LOD by
scientific organizations. Methods of calculating LOD are summarized
and the results compared. Current regulatory limits of detection
for two EPA programs for hazardous substances are then tabulated and
discussed. The substances listed are those likely to be present at
release sites, based on reported findings. Following this, an
assessment is made as to whether the operational LODs for hazardous
substances are likely to be lowered significantly in the near term
(within the next five years). Data from two programs (the
Association of Official Analytical Chemists and the New York State
Department of Health), where many laboratories analyzed the same
samples, are summarized. Evidence is presented for a general
-------
barrier to LODs in the low-ppb concentrations. Available evidence
is assessed as showing little or no near-term LOD improvement.
LOD for radiochemical measurement of nuclides reported at
release sites are tabulated. Calculated LODs and values obtained in
a commercial laboratory are compared to each other. LODs for the
more usual methods of chemical analysis are compared with LOD values
for counting methods.
-------
2.0 ORIGIN AND DEFINITIONS OF LOD
2.1 Origin of the LOD Concept
Historically, the concept of the chemical LOD as a
specification for an analytical chemistry procedure was first
developed in the 19th century. The intended use of laboratory LODs
was to indicate that a certain analytical method could measure the
amount of a given chemical within a specified concentration range.
The early analytical methods had LODs in the concentration range of
a tenth of a percent (ppt) to parts-per-million (ppm), expressed
usually on a weight-to-weight or volume-to-volume basis. Then, in
the 20th century, analytical research development lowered the
then-attainable LOD values down to the parts-per-billion (pp,b) and
even parts-per-trillion (ppt) levels. This decrease was largely due
to the introduction of new technologies into the analytical
laboratory.
Scientific interest in improvement and refinement of
expressions of LOD continues up to the present time. A recent
symposium, held as part of the 191st National Meeting of the
American Chemical Society (ACS) in April 1986, examined both
scientific and regulatory/legal aspects of the specification of
LODs. In this two-day symposium, over half the papers dealt with
legal or regulatory LOD concerns. Lowering of LODs, more
reliable specifications for LOD, and consequences of using two
procedures to measure one substance, which then yields two different
-------
LODs are among topics addressed in the symposium papers. What is
new about this meeting is that the analytical chemists are
addressing both scientific and legal utilization of LOD data.
2.2 Description of the LOD
The LOD has always been an important numerical expression in
the field of chemical analysis because it is a way to communicate,
from one analyst to another, the lowest concentration of a specified
chemical which can be measured using a given analytical method. The
ACS's 1980 definition of LOD states that "the limit of detection is
the lowest concentration of an analyte that an analytical process
..(2)
can reliably detect."
LOD is an operational parameter, not an invariant physical
constant, for any analytical procedure or for any substance. (The
molecular weight of a chemical is, for example, a physical
constant.) LOD is an operational parameter because it depends on a
variety of factors which vary slightly every time the analytical
procedure is performed on a group of samples.
The usual way of carrying out a trace chemical analysis
consists of two distinct unit operations:
• A sample-workup operation, in which the chemical to be
quantified is extracted from most of the other substances in
the sample matrix. Usually some concentration and "cleanup"
of the extract are also performed to enhance the analytical
signal of the chemical above the matrix background noise.
• An instrumental-analysis operation, in which the chemical in
the "cleaned-up" extract is quantified by an instrument
containing a detector. The instrument typically converts
10
-------
the quantity of chemical present in the detector at any time
to a proportionate electrical signal, which is, in turn,
manipulated by on-board microprocessors to yield the best
signal (measurement) possible. Background interference
caused by the sample matrix can decrease the net measurable
signal from the chemical, and thereby raise the LOD.
The operational LOD obtained is thus affected by each of the above
operations. One can improve or degrade the LOD by varying the
quality of either the sample-workup or instrumental-analysis steps.
Day-to-day LOD variation depends on such variables as solvent
quality and instrument stability. This means that even though the
whole procedure may have a fairly constant LOD on the average, the
daily LOD for the procedure will vary somewhat. The size of.this
variation depends on the individual procedure. Application of one
procedure to two different sample matrices usually leads to
i
different LOD values for each matrix.
2.3 Calculation of the LOD
A further cause of an apparent variation in LOD values is that
there is more than one acceptable way to calculate an operational
i
LOD from one single set of experimental data using one of several
acceptable calculation techniques. The variation in LOD which can
result from different methods of calculation is illustrated by the
(3)
work of Long and Winefordner. These investigators calculated
three LOD results for calcium using one set of experimental data
(from measurements of calcium concentrations in water using plasma
(3)
spectroscopy). Each of the three mathematical equations uses
11
-------
different pieces of information taken from the same instrumental
calibration line to yield different LODs. Following are three of
the methods used for the calculations:
• A mathematical statement of the International Union of Pure
and Applied Chemistry (IUPAC) definition of LOD.
• A graphical definition which takes into consideration the
standard deviation of the slope of the calibration line.
• A procedure that includes the errors in both the intercept
and the slope of the straight-line calibration plot.
Using the same spectroscopic data to construct one calibration line,
Long and Winefordner arrived at three different LODs for calcium by
following three different algorithms. The values are 0.002, 0.003,
and 0.05 ppm. The different values result from different mathe-
matical equations used to express the LOD. The results vary by over
an order of magnitude even though portions of the same experimental
data were used in all three calculations.
12
-------
3.0 REGULATORY SPECIFICATION OF LODs
Regulatory LODs can be used to specify how well an analytical
method (or more correctly, a laboratory using that method) must
perform to be acceptable for use in a particular regulatory
(4)
program. Therefore, laboratory LODs should always be lower in
concentration than regulatory LODs. Table 1 lists the LOD data
specified in two EPA programs for a variety of substances. The
substances listed in Table 1 are those most frequently reported at
nearly 900 sites listed or proposed for inclusion on the National
Priorities List (NPL). The LOD specifications are taken from the
EPA Contract Lab Program (CLP)^ ' )7' and also from the Clean
(8)
Water Act regulations . The respective LOD requirement from
each program is paired with the frequently occurring substances from
the NPL sites, where such regulatory LOD data currently exist.
Some observations can be made about the data collected in
Table 1:
• For some hazardous substances frequently found at NPL sites,
there are no LOD specifications listed for either EPA pro-
gram (e.g., ethylbenzene, PCBs, benzo(j,k)fluorene, uranium).
• The LOD criteria set by the two programs for the same
substance are frequently different (e.g., trichloroethylene,
chloroform, and 1,1-dichloroethane).
• The LOD specifications for the Clean Water Act program are
variable, whereas the CLP numerical limits tend to be more
uniform. LOD values of 5, 10, and 330 ppb can be seen
regularly in columns 3 and 4 (CLP program). LOD data for
CLP have clearly been collected for a variety of chemicals
and then this array of data has been "rounded off" to a
smaller number of expected LOD values.
13
-------
TABLE 1
LIMITS OF DETECTION REQUIRED BY TWO EPA PROGRAMS FOR SUBSTANCES FOUND
AT NPL SITES (SUBSTANCES ARE LISTED BY FREQUENCY OF REPORTED OCCURRENCE)
Substance Name
Trichloroethylene (TCE)
Lead (Pb)
Toluene
Chromium and Compounds,
NOSe (Cr)
Benzene
Chloroform
Polychlorinated Biphenyls,
NOS (PCBs)
1, 1, 1-Trichloroethane
1, 1, 2,2-Tetrachloroethene
Zinc and Compounds, NOS (Zn)
Cadmium (Cd)
Arsenic
Phenol
Xylenes - Total
Ethylbenzene
Copper and Compounds, NOS (Cu)
1, 2-Trans-Dichloroethene
Methylene Chloride
1 , 1-D ichloroethane
1,1-Dichloroethylene
Mercury (Hg) NOS
Cyanides (Soluble Salts), NOS
Vinylchloride
Nickel and Compounds, NOS (Ni)
Chlorobenzene
1, 2-Dichloroethane
Carbon Tetrachloride
Pentachlorophenol (PCP)
Naphthalene
Methyl Ethyl Ketone
Trichloroethane, NOS
Site
Frequency3
311
286
243
220
208
179
159
151
149
142
141
141
121
113
111
106
104
91
85
79
78
73
70
65
64
64
61
53
48
42
38
Required
Limits of Detection
CLPb
^gfr
5
5
5
10
5
5
—
—
—
20
5
10
10
5
—
25
—
5
5
5
0.2
10
10
40
5
5
5
50
10
10
5
CLPC
^JTkg
5
—
5
—
5
5
—
—
—
—
—
—
330
5
—
—
—
5
5
5
—
—
10
—
5
5
5
1600
330
10
5
CWAd
ug/L
0.12
—
—
—
—
0.05
—
0.03
0.03
—
—
—
—
—
—
—
0.10
0.25
0.07
0.13
—
—
0.18
—
0.25
0.03
0.12
—
—
—
0.03
14
-------
TABLE 1 (Continued)
Substance Name
Site
Frequency3
Required
Limits of Detection
CLPb
CLPC
CWAd
Iron and Compounds, NOS (Fe) 33
Barium (Ba) 32
Manganese and Compounds, NOS (Mn) 31
Acetone 30
Phenanthrene 28
Benzo[a]Pyrene 27
1,1,2-Trichloroethane 25
Anthracene 22
Styrene 22
DDT 22
Tetrachloroethane, NOS 21
Lindane 21
Bis(2-Ethylhexyl)phthalate 21
Selenium 20
Sulfuric Acid 20
Pyrene 19
Aluminum and Compounds, NOS (Al) 18
Fluorene 17
Benzo(j,k)Fluorene 17
Ethyl Chloride 16
Radium and Compounds, NOS (Ra) 168
Dichloroethane, NOS 15
Trichloroflouromethane 15
Acenaphthene 14
Uranium and Compounds, NOS (U) 148
cis-l,2-Dichloroethylene 14
Trinitrotoluene (TNT) 13
Asbestos 13
Antimony and Compounds, NOS (Sb) 13
Chlordane 13
Dichlorobenzene, NOS 12
Radon, NOS (Rn) 128
Hexachlorobenzene 12
Tribromomethane , 12
Ammonia, NOS 11
60
0.5
100
200
15-
10
10
10
5
10
0.1
5
0.05
10
5
10
200
10
10
5
10
—
—
—
10
330
330
5
330
16
5
8
330
—
330
—
330
10
5
330
0.02
80
0.52
0.03
0.20
15
-------
TABLE 1 (Continued)
Site
Substance Name Frequency3
Chloromethane
Di-n-Butylphthalate
Tetrahydrof uran( I)
Thorium and Compounds, NOS (Th)
DDE
Dioxin
Dibromochloromethane
Methyl Isobutyl Ketone
(4 methyl-2-pentanone)
Chrysene
Dieldrin
1,4-Dichlorobenzene
2,4-Dinitrotoluene
Hexachlorocyclopentadiene (C56)
Cresols
Beryllium and Compounds, NOS (Be)
Nitrates, NOS
Endrin
2, 6-Dinitrotoluene
Ethyl Ether
1, 2-Dichlorobenzene
Diethylphthalate
Bromomethane
Fluoride (ion), NOS
Diethylphthalate
Cyclonite (RDX)
Aldrin
Boron and Compounds, NOS (B)
Heptachlor
Hexachloro butadiene (C46)
1, 2-Dichloropropane
Endrin
Silver, NOS
1,2,3-Trichloropropane
Chlorodifluoromethane
Chloride (ion) NOS
11
11
11
108
10
10
10
10
10
9
9
9
9
9
8
8
8
8
8
8
8
8
8
8
8
7
7
7
7
7
7
6
6
6
6
Required
Limits of Detection
CLPb
ug/L
10
10
—
—
0.1
— f
5
10
10
0.1
10
10
—
—
5
—
0.1
10
—
10
10
10
—
10
—
0.05
—
0.05
10
5
0.01
10
—
—
—
CLPC
ugTkg
10
330
—
—
16
f
5
10
330
16
330
330
—
—
--
—
16
330
—
330
330
10
—
330
—
8
—
8
330
5
16
—
—
—
—
CWAd
iiZL
0.08
—
—
—
—
—
0.09
—
—
—
0.24
—
—
—
—
—
—
—
—
0.15
—
1.18
—
— -
—
—
—
—
—
0.04
—
—
—
1.81
—
16
-------
TABLE 1 (Concluded)
Site Required
Substance Name Frequency3 Limits of Detection
ODD
Dibenzofuran
Cobalt and Compounds, NOS (Co)
Sulfate (ion), NOS
Methane
Acenaphthylene
Hexachlorocyclohexane, NOS
1,2,4-Trichlorobenzene
Trimethyl Benzene
Benzo [a Janthracene
2-Chlorophenol
2, 4-Dimethylphenol
Hydrogen Sulfide
Methanol
6
6
5
5
5
5
5
5
5
5
5
5
5
5
CLPb
ug7L
0.1
10
50
—
10
—
10
—
10
10
—
—
"
CLPC
ujTkg
16
330
—
—
—
330
—
330
—
330
330
—
"
CWAd
ug/L
—
—
—
—
—
—
—
—
—
—
—
—
—
"
aNumber of sites at which a chemical has been identified through NPL
Update 5 (888 sites).
bContract Required Detection Limits (CRDL) for water: U.S. EPA Contract
Laboratory Program, Statement of Work for Organics Analysis, Revised
January 1985; and Inorganic Analysis, July 1984.
CCRDL for soils and sediment: U.S. EPA Statement of Work for Organics
Analysis, op.cit.
^Method Detection Limits in water: taken from Federal Register/Vol. 49,
No. 2091, Friday, October 26, 1984; 40 CFR Part 136, Guidelines
Establishing Test Procedures for the Analysis of Pollutants Under the
Clean Water Act, Table 1, p. 43265.
eNOS = Not otherwise specified as to isomers, oxidation state, counter
ions.
^Performance criteria for the GC/MS method for Dioxin (TCDD) described in
the Statement of Work, Dioxin Analysis, U.S. EPA Contract Laboratory
Program, September 1983.
SRadioactive element. See Table 2 for LOD data.
17
-------
The above observations reflect the somewhat subjective
procedures that the two programs use in setting the regulatory LOD
values shown in Table 1. The programs have different objectives
and, consequently, the value of the LOD required for a given
chemical in a given program may be different.
18
-------
4.0 POTENTIAL FOR IMPROVEMENTS IN LABORATORY LOD VALUES
This section develops a forecast of near-term (next five years)
major improvements in LOD values for the chemical analysis of
environmental samples. This forecast is based first on an exami-
nation of past times when LODs were improved dramatically and then,
on an identification of the causes of that improvement. A second
basis for forecasting LOD improvements is an evaluation of current
barriers which are seen as preventing substantial LOD improvement at
the present time. A conclusion is drawn that LODs in environmental
laboratory procedure will not improve significantly over the near
term.
4.1 Past Improvements in the LOD
In the past, improvements in the best obtainable LODs have come
about in one of two ways:
• Through gradual improvements in the quality of reagents
employed or in the more stable operation of instruments,
which have led to incremental improvements in LOD values
• Through dramatic improvement in reagent quality or as a
result of invention of a new instrument or detector which
has led to sudden decreases in LOD (over several orders of
magnitude)
The time period when the most dramatic improvements occurred in
analytical chemical LODs was the period 1945-1970. During this time
frame, an abundance of new electronic technology and equipment de-
rived from World War II programs became available. Also, personnel
trained during the war in the use of these electronic components
19
-------
began to work in analytical chemistry research laboratories. The
combination of new technology and trained people resulted in new,
sensitive, chemical instrumentation such as: atomic absorption
spectroscopy, mass spectrometry, gas chromatography, the electron
capture detector, and the ubiquitous application of high-speed,
digital mini-computers in chromatography. Detection limits, in
general, decreased to ppt levels or lower during this period.
However, two sources of error became important when detection
limits reached the ppt level: (l) adsorption on laboratory
apparatus; and (2) trace impurities in reagents and solvents. At
this level, adsorption to sites in the field sample containers and
laboratory glassware, as well as in the instruments, resulted in
losses of significant fractions of substances originally present in
(9)
the sample, thus making detection in the ppt range difficult.
For example, the losses of a chemical during the sample-workup
resulted in extracts containing a lower amount of chemical than the
original sample from which the extracts were derived. Also, traces
of impurities in solvents and reagents became detectable by the
instruments when they were operated in the ppt range. Signals from
these impurities sometimes obscured the signal caused by the sample,
resulting in an LOD at only high ppt levels. These phenomena
(chemical adsorption and reagent impurity) have created an impasse.
Ways to easily overcome adsorption losses, such as coating
(9)
glassware, have not been found; and substantially higher purity
20
-------
reagents and solvents have not been developed (by distillation,
etc.) despite attempts to do both. Strides in instrumentation have
(9)
not been able to overcome these effects.
4.2 LOP Limits from Interlaboratory Data
Recent evidence has been published which indicates that poor
precision has been encountered for a variety of analytical methods
at concentrations below 10 ppb. Generally, a loss of
precision makes measurements at low concentrations more difficult,
and the LOD reaches a limit. Based on a statistical study of
precision data from 18 separate Association of Official Analytical
Chemists (AOAC) interlaboratory studies, precision seriously
deteriorates below 10 ppb. In an interlaboratory study, all
laboratories conduct an analysis of the same samples using the
identical analytical procedure. Each laboratory submits its answer
to the AOAC for statistical analysis. For each of the 18 studies, a
different analytical procedure was used and a different substance
was measured.
The numerical value used to measure precision in the analysis
of the 18 studies was the among-laboratories relative standard
deviation, RSD . The RSD is calculated by first computing the
X X
standard deviation for all* the results submitted by the
laboratories. This standard deviation is then divided by the mean
*Statistical outliers are removed first.
21
-------
of results submitted by all laboratories and multiplied by 100 to
yield RSDx, in units of percent. A large numerical value of
precision means there is poor agreement among participants, and a
small value of RSD means there is close agreement in results
A
among laboratories.
The among-laboratories RSD , as a function of sample
concentration, C, was found (for all procedures used in all 18
studies) to increase with decreasing C according to the following
(9)
equation :
RSDX = 2^ ~ °'5 lQS C)
This means that the RSD increases exponentially as the
X
concentration being measured decreases. In only 3 of 18 studies
(9)
reviewed was there any deviation from the above equation. The
deviation from the equation in the three studies was found to be
caused by the inability of the instrument used to respond to
chemical concentrations less than 100 ppm.
Even for methods which gave a good instrument signal down to
1 ppb, the same relentless decay in precision was encountered as
sample concentrations decreased. The descriptive equation that
gives the relative error as a function of concentration, and the
comparative laboratory data themselves, show that the RSD is
usually about +12 percent at sample concentrations of 10 ppm and
that RSD more than doubles to +30 percent at a concentration of
10 ppb. Because the RSD is an exponential function of
22
-------
sample concentration, it will increase in magnitude rapidly for
measurements of concentrations below the 1 ppb level.
The precision versus concentration studies assembled to date
(9)
from AOAC data are extensive, but not exhaustive. The evidence
demonstrates that operational LODs at sample concentrations below
1 ppb are not imminent due to the exponential increase in random
error at such low concentrations. Without some revolutionary
breakthrough in technology that provides a practical means of
improving attainable precision at low concentrations, the LODs will
not improve substantially.
4.3 LOP Limits From Proficiency Testing Data
Further evidence that current LODs may not decrease
dramatically in the near term comes from studies conducted by the
New York State Department of Health, which conducts a quarterly
proficiency testing activity as part of their laboratory
certification program. In proficiency testing, separate
portions of one master sample are sent to participating laboratories
and all the portions are analyzed at about the same time.
Ninety chemistry laboratories participated in recent New York
State proficiency testing, and the results are in accord with the
(9)
previously described study of collaborative analyses. Seventy
different test solutions of chemicals were analyzed by the parti-
cipating laboratories. The best reported precision was +8 percent
(two relative standard deviation units). This was obtained with the
23
-------
highest concentration sample, a 1.51 ppm solution of copper.
Conversely, the worst precision, +69 percent, was reported for the
lowest concentration sample in the program. This sample was a 0.14
ppb solution of the pesticide Silvex.
Once again, precision is seen to deteriorate as the sample
concentration was lowered, making it increasingly difficult to
detect the substance of interest with any certainty. Data for the
test results for other chemicals used in the New York State
Proficiency Testing Program also support the observed decrease of
precision with decreased concentration for all methods.
LODs will not improve appreciably until the exponential
increase in uncertainties in analytical measurements at concen-
trations less than 1 ppb can be overcome by new technology.
24
-------
5.0 EQUIVALENCE OF LCDs FOR RADIOCHEMICAL COUNTING COMPARED TO
SPECTROSCOPIC METHODS.
5.1 Selection of Elements to be Considered
For radioactive elements there is additional concern that
current radiochemical detection limits may be significantly smaller
than those for more conventional techniques. If so, it is con-
ceivable that radioactive substances identified in environmental
samples by radiochemical means might not have been detected if other
methods of chemical analysis were used. Since the HRS ranking
factor for toxicity is based on chemical hazard, not radiation
hazard, radiochemical detection could bias the relative ranking of
sites containing radioactive elements. In this section, typical LOD
values for radiochemical techniques are compared with LODs for the
spectroscopic methods used for CLP procedures. The CLP methods use
inductively coupled plasma emission spectroscopy and atomic absorp-
tion spectroscopy to determine the inorganic elements included in
the program.
There are four radioactive elements listed in Table 1: Radium,
Uranium, Radon, and Thorium. The LOD values for these will be
considered here. A fifth nuclide, Plutonium, has been included in
the present discussion because of special interest in detecting
small amounts of this element. This section addresses the question
of what the LODs would be, in the same units as used in Table 1, if
radiochemical counting methods were used to quantify these five
radioactive elements found at hazardous substance release sites.
25
-------
5.2 Quantitative Principles for Radionuclides
In order to quantify a nuclide by radiochemical counting, two
things are required. The first is an instrument capable of
detecting and counting the radiation given off by the nuclide of
interest; and the second is a sample nuclide with a reasonable
half-life. A reasonable half-life may be defined as one short
enough to give adequate count rates, and yet long enough so that the
concentration of the isotope does not change significantly during
the time frame of the measurement.
Radiochemical methods have the potential for low LODs because
it takes only one atom to give one particle emission, which in turn
can give one count at the instrument. The one-atom, one-count
relationship is inherently very sensitive, given the large number of
atoms present even at trace concentrations. However, just as for
other instrumental methods of analysis, one or more factors
affecting steps in the procedure can degrade the operational LOD
dramatically. For example, the counter detector can be less than
100 percent effective at counting the emitted radiation. This loss
of efficiency can be due to counting geometry, detector or
electronic overload. Absorption of emissions by any matrix
material, or by changes in geometry between sample and detector, can
occur. Such matrix effects lower the observed count rates for a
given level of radioactivity, and thus the effects degrade the
operational LOD for that sample or mixture.
26
-------
If the radiochemical analysis method consists of unit
operations of sample workup and instrument measurement (such as
described earlier), then either or both operations can affect the
LOD obtained on any given day. Therefore, radiochemical analysis
then becomes similar to other instrumental methods of analysis,
except that the instrument signal is triggered by radiation from the
isotope being measured. Thus, any loss of the nuclide in the sample
workup operation will unfavorably affect the LOD of a radiochemical
method.
5.3 Inorganic LODs for Comparison Purposes
In order to compare radiochemical measurements with other
techniques, some reference point data are needed. A set of LOD data
for metals in environmental samples has been established as part of
the CLP. The LODs for 24 metals in water range from 0.2 to
5000 ug/L, as measured by inductively coupled plasma emission or
atomic absorption spectroscopy. Eighteen of the 24 LODs are
specified to be 100 ug/L or less. So, for a range of comparison,
0. 5 to 100 ug/L would be reasonably representative of the CLP
expected performance of spectroscopic methods used to measure
18 different metals. This range will be used for comparisons with
radiochemical LOD ranges, because radioactive element LODs are not
specified in the CLP program at this time.
5.4 LOD Values for Radioactive Elements in Environmental Samples
A search proved unsuccessful in providing literature which
addressed the exact environmental problem under consideration.
27
-------
Alternative sources of LOD data were located and contacted. Two
independent means of establishing LOD values were identified and
used. First, an EPA research scientist in the Office of Radiation
Programs calculated LODs from properties of the radionuclides
(13)
selected and current counting technology. Independently, the
director of a commercial environmental testing laboratory
(specializing in radioactive tests) provided the operating LOD
(14)
values currently in use at their facility. Neither of the two
sources of data were aware of the other.
Table 2 summarizes both calculated and operational LODs for the
five radioactive elements of interest. ' Half-life data are
also tabulated. For the first two radionuclides, Uranium and
Thorium, both the calculated and experimental LODs are around 1 ug/L
(approximately 1 ppb) . These LOD values are in the range of the
spectroscopic LODs for many nonradioactive metals in CLP samples, as
presented in Section 5.3.
In contrast, the LODs for radiochemical quantification of both
—7 —fi
Radium and Plutonium are on the order of 10 to 10 ppb and
are therefore a million-fold smaller than any LOD in the current CLP
/ f \
listing. The relatively short half-lives of Ra and
239
Pu (1,600 and 24,390 years, respectively) cause high counting
rates and therefore low LOD values.
One single value of the LOD for Radon gas cannot be listed in
Table 2 because the methods of trapping and collecting the Rn gas
28
-------
from the rest of the sample matrix are important determinants of the
LOD finally achieved.
For radiochemical LODs, therefore, both counting instrument
sensitivity and isotope half-life play an important role in
determining how low a concentration can be detected. In general,
the LOD can be expected to decrease for isotopes with shorter
half-lives, as listed in Table 2.
29
-------
TABLE 2
LIMITS OF DETECTION FOR
RADIOACTIVE SUBSTANCES FOUND AT NPL SITES
Element, Isotope3
Th, 232
U,238
Pu,239
Ra,226
Rn,222
Isotope
Half-Life
1. 4lxl010year
4.51xl09year
24, 390year
1, 600year
3. 82day
Limits of
Calculated13
1.0 ug/L
0.3 ug/L
1. 5xlO~6ug/L
lxlO~7ug/L
(gas)d
Detection
Experimental0
5.4 ug/L
1.8 ug/L
0. lxlO~6ug/L
0. 6xlO~6ug/L
(gas)e
a For each element a minimum instrument count rate is required to
measure the isotope. The instrument limit assumed for calculation of
LODs is 0.1 pCi/L. The experimental count limit is given as 0.6 pCi/L
for LOD measurements. A Curie (Ci) is 3.7 x 10^0 disintegrations
per second.
Personal communication, Dr. Paul Hahn, EPA Office of Radiation
Programs, Las Vega
for Rn gas in air.
Programs, Las Vegas, NV. Units are for ground water matrix, except
c Personal communication, James Mueller, Controls for Environmental
Pollution, Inc., Santa Fe, NM. Units are for ground water matrix,
except Rn'gas.
** LOD for Radon depends on method used to collect (and thus
concentrate) gas.
e Operational detection limit given as 0.1 pCi/m3 based on 400 m3
of air collected, d^)
30
-------
6.0 CONCLUSIONS
6.1 Anticipated Changes in Operational LODs
The numerical value of an operational laboratory LOD obtained
experimentally will depend on the combination of equipment and
reagents involved, the sample throughput rate and the sample
concentration, and the calculation algorithm. The operational LOD
obtained in the laboratory on a day-to-day basis further depends on
the care expended in doing each step as well as the skill and
experience of the analyst doing the work. At the present time,
there is no apparent prospect that dramatic improvements in current
sample processing and instrument technologies will be able to
improve the LODs, much beyond those values listed in Tables 1 and 2,
within the next five years. The different values of LOD specified
by different regulatory programs for the same compound, as given in
Table 1, show that some programs may require something less than the
state-of-the-art LODs to achieve a regulatory goal. It is not clear
that future improvements in laboratory LODs would necessarily lead
to corresponding changes in regulatory LOD values, because the
present values may well be adequate to accomplish the regulatory
objective.
Additionally, any new technology must do two things before
being integrated into current environmental laboratory operations.
First, the new technology must demonstrate its ability to provide
reliable, low LOD values in the matrices of the common environmental
31
-------
samples. Second, the new technology must meet the requirements of
standardized environmental laboratory programs such as the CLP.
These requirements Include the use of proven, standard methods, and
the availability of affordable equipment, supplies and repair ser-
vices. A new technology offering a lower LOD without also meeting
these other requirements will not be sufficient. It is necessary to
demonstrate that the technology is workable by providing substantial
LOD improvement in the work environment of a for-profit testing and
production laboratory.
The most direct way to obtain a short-term, immediate
improvement in an LOD is to decrease the size of the statistical
variability (i.e., improve precision) in the analytical data for a
given chemical. This improvement can result from incremental
improvements in the quality of the instrumental and/or the sample
extraction and concentration steps. It can also be improved by
increasing the number of replicate analyses run on the same sample.
Both steps to improve method precision and thus decrease the LOD
always increa'se cost-per-sample-analyzed substantially, and are
therefore not in general use in environmental analytical
laboratories.
In summary, a major "breakthrough" in the lowering of LODs will
only occur if a completely new instrument (capable of analyzing
components in environmental samples) is introduced or if a new
sample workup technique should be developed. The last few years
32
-------
have not produced major breakthroughs in either environmental
instrumentation or methodology. Improvement of LODs for existing
methodologies (such as used in the CLP program) over the next
five years should therefore be gradual, if at all.
6.2 LOP Comparisons Between Radiochemical and Spectroscopic Techniques
Radioactive metals, reported present at a few NPL sites, do not
have LOD values established as part of the CLP program at the
present time. For purposes of comparison, an estimate of LODs which
might be expected for metals can be obtained by examination of CLP
limits for metals determined by atomic spectroscopy. The current
CLP program sets limits between 0.2 and 100 ug/L for 18 of the
24 nonradioactive elements measured by atomic spectroscopy. This
range of values was compared with the LODs for five naturally
radioactive elements measured by radiochemical counting.
LOD specifications using radiochemical counting techniques
depend on the radiochemical decay rate of the isotope being
measured. The observed decay rate of a given isotope is a function
of the half-life which is a physical constant, and is invariant for
the particular nuclide being measured. In general, the shorter the
half-life of an isotope, the smaller will be the LOD of that
element. Therefore, based on Table 2, for elements with longer
half-lifes (e.g. Thorium, Uranium), LODs for radiochemical and
spectroscopic methods are generally comparable. LODs for
radiochemical counting are much lower for those elements with
shorter half-lifes (e.g., Radium and Plutonium).
33
-------
REFERENCES
1. Abstracts of the 191st National Meeting of the American
Chemical Society, New York, N.Y., April 13-18, 1986. Division
of Analytical Chemistry, Abstracts 54-58, 71-77, 91-96, and
108-111.
2. L.H. Keith, et al., "Principles of Environmental Analysis,"
Anal. Chem., 55, pp.2210-2218, 1983.
3. G.L. Long and J.D. Winefordner, "Limit of Detection, A Closer
, Look at the IUPAC Definition," Anal. Chem., 55, pp. 712A-724A,
1983. ~
4. L.B. Rogers, et al., Eds., "Recommendations for" Improving the
Reliability and Acceptability of Analytical Chemistry Data Used
for Public Purposes," C&E News, j>0, (23)44, pp.44-48, 1982.
5. U.S. EPA Superfund Contract Laboratory Program Statement of
Work for Organics Analysis," Exhibit C, Revised January 1985.
6. U.S. EPA Superfund Contract Laboratory Program Statement of
Work for Inorganic Analysis, SOW No. 784, Exhibit C, July 1984.
7. U.S. EPA Statement of Work, Dioxin Analysis Soil/Sediment
Matrix Multi-Concentration, September 15, 1983.
8. Fed. Reg. 49, No. 2091, 40CFR, Part 136, "Guidelines
Establishing Test Procedures for Analysis of Pollutants Under
the Clean Water Act," Table 1, p.43265ff, Fri., Oct. 26, 1984.
9. L.B. Rogers, "Problems with Interlaboratory Measurements at
Trace Levels" Abstracts of 191st National Meeting, American
Chemical Society, Abstract ANYL 55, April 1986.
10. K.W. Boyer, W. Horwitz, and R. Albert, "Interlaboratory
Variability in Trace Element Analysis," Anal. Chem., 57,
pp.454-459, 1985.
11. J.C. Daly and K.E. Asmus, "Laboratory Performance in
Proficiency Testing," Environ. Sci. Technol., 19, pp.8-13, 1985.
12. H.H. Willard, L.L. Merritt, and J.A. Dean, "Instrumental
Methods of Analysis," Chap. 11, Radiochemical Methods, pp.
302-327, 5th ed., 1965.
35
-------
REFERENCES (Concluded)
13. Personal Communication: Dr. Paul Hahn, EPA Office of Radiation
Programs, Las Vegas, NV, July 1985.
14. Personal Communication: James Mueller, Pres., Controls for
Environmental Pollution, Inc., Santa Fe, N.M., March 1986.
36
------- |