|
YUE RONG, Ph.D.
California Regional Water Quality Control Board – Los Angeles
Region
320 West 4th Street, Suite 200, Los Angeles, CA 90013
Tel: (213) 576-6710, E-mail:
yrong@rb4.swrcb.ca.gov
(December 31, 2001)
(Revised February 1, 2002)
What about laboratory detection
limits? In environmental
forensics studies, decisions are often based on analytical
data indicating certain contaminants as being “detected” or
“non-detect.” The detection result is of course based on the
“detection limit,” a conceptually easy but technically
complicated term. Since detection limits are analytical method
specific, we must first review the concepts and definitions
associated with analytical method systems and specifications.
In this article, the analysis of volatile organic compounds (VOCs)
is used as an example to discuss this subject.
To detect VOCs, gas chromatography
(GC) is used to separate the volatile compounds based on their
volatility and instrumental column phase affinity. Basic
components of a GC include sample injection port, helium
carrier gas tank, heated oven, within which there is a gas
separation column, and gas exit with connection to a detector
(Figure 1) (Pavia et al. 1995). Environmental sample is being
injected through GC and detected by a particular detector.
Various detectors are used to identify and measure the
compounds. Commonly employed detectors for VOCs include flame
ionization detector (FID), photo ionization detector (PID),
electron capture detector (ECD), electrolytic conductivity
detector (ELCD, also known as Hall detector), and mass
spectrometer (MS). FID uses flame to burn eluting compounds
and then measures ions to detect the compounds. FID is best
suited to detect combustible compounds such as gasoline and
methane. PID uses ultraviolet radiation to ionize target
compounds and then measure resulting ions to detect the
compounds. PID is sensitive to compounds with double bonds,
such as aromatic compounds and ethylene. ECD uses the
principle of electron absorption to measure the electrons to
recognize and detect target compounds. ECD can be used to
detect almost all volatile or semi-volatile compounds. ELCD
uses high temperature to flush out halogen ions from the
target compounds and then measure the conductivity to
determine the compounds. ELCD is very sensitive to halogenated
compounds, such as tetrachloroethylene (PCE) and
trichloroethylene (TCE). MS uses electronic energy to bombard
the target compounds and then measure the fragmentation
pattern to determine the identity of the compounds. Since the
fragmentation pattern is compound specific, MS can most
accurately identify an unknown compound. FID, ELCD, and MS are
destructive detectors to target compounds, while the PID is
non-destructive. As far
as the sensitivity to detection limit relative to each
detector, FID does not generally have a very low detection
limit. In comparison with FID, PID has relatively a low
detection limit for compounds with double bonds. Detection
limit for ECD or ELCD is improved with lower levels in
comparison with FID and PID. Detection limit associated with
ELCD is lower than ECD particularly for halogenated compounds.
MS has most accurate ability to identify an unknown compound,
but has the relatively high detection limit comparing with all
other above-mentioned detectors. Note that the comparisons of
detection limits among the detectors are relative to target
compounds. The comparison is sometime difficult because each
detector is designed to detect certain group of compounds, and
therefore sensitive to that group of compounds with better
detection limit. Given these different types of detectors, the
U.S. Environmental Protection Agency (EPA) has specified
certain analytical methods. For example, EPA Methods 502.1 and
601 are GC/ELCD method for halogenated compounds; EPA Methods
503.1 and 602 are GC/PID method for aromatic compounds; EPA
Methods 502.2 and 8021 are GC/PID+ELCD/FID method for both
aromatic and halogenated compounds; and EPA Methods 524.2,
624, and 8260 are GC/MS method for aromatic and halogenated
compounds.
Analytical chemistry has two
important aspects: (1) to identify the unknown compounds (also
called qualification) and (2) to measure concentration of the
compounds (also called quantification). The identification is
completed by using the retention time, which is the time used
for volatile compounds to travel through the GC separation
column. The GC column is sensitive enough to separate volatile
organic compounds by their volatility. In principle, volatile
compounds with higher volatility and less affinity to the
column will come through the column and arrive at the detector
faster than those with lower volatility compounds. By counting
the retention time on the chromatogram, the compounds can be
differentiated. The quantification is completed by comparing
the unknown environmental sample concentration with the known
standard concentration. As the detector relates given standard
concentration to a certain amount of mass response in the
instrument, the mass amount will then be related
proportionally to unknown sample concentration. The concept of
detection limit primarily relates to the quantification aspect
of the analytical method.
The detector provides the
electronic signals to register the mass to determine the
concentration. The lower the concentration is, the more
difficult the signals can be clearly distinguishable from
other electronic background noise. Therefore, conceptually
there has to be some limit, below which we cannot say if there
is detection or not. In that concept, scientists try to bring
in some numerical standard to specify the limit. One important
concept is called the “method detection limit” (MDL). MDL is
defined as the minimum concentration of a substance that can
be measured and reported with 99% confidence that the analyte
concentration is greater than zero. Per the Code of Federal
Regulations (CFR) (CFR 40, part 136, Appendix B, 1993), the
practical protocol to determine MDL specifies mathematically
to take a minimum of 7 replicates of given spiking
concentration in a range of one to five times from the
projected lowest concentration that detector in the analytical
method can measure, and then calculate with the following
equation: MDL = SD x
t0.99 (1)
Where SD = {ĺni=1(xi
–X)2
/ (n-1)}˝ MDL
= method detection limit (mg/L),
SD = standard deviation,
t0.99 = t-distribution table value for 99% with the
degree of freedom (n-1),
xi = spiking replicates concentration (mg/L
) (i = 1 . . . n) (n=7 in this case),
X = the mean of spiking
concentrations (mg/L).
As we can see, the so-called 99%
confidence is really based on the t-distribution in
statistics. Of course, this assumes that the distribution of
the low level spiking concentrations follows the
t-distribution. For
example, there is a real case of MDL study for methyl tertiary
butyl ether (MTBE) using EPA Method 524.2 by a commercial
laboratory. The seven replicates spiking concentrations (mg/L)
are 0.45, 0.46, 0.49, 0.46, 0.45, 0.50, and 0.53,
respectively. Based on these data, we have X = 0.48 and SD =
0.0304. Given t0.99 (n=7) = 3.143,
Equation (1) gives MDL = 0.0955. Therefore, this MDL study
produces the method detection limit (MDL) for EPA Method 524.2
is about 0.1
mg/L
for MTBE. Along with the
definition of MDL, other concepts and terms of detection
limits can also be derived. Instrument detection limit (IDL)
is the lowest detection above equipment electronic signal
noise, which most likely below MDL. Although MDL study can
produce a low level for detection, most commercial analytical
laboratories are still not very confident to quantify the
concentration detected slightly above the MDL due to many
other laboratories and environmental sampling uncertainties.
Therefore, the practical quantitation limit (PQL) and
estimated quantitation limit (EQL), used interchangeably
sometime, are set to increase the confidence level in
quantification. PQL or EQL is defined as 2 to 10 times above
MDL. By raising MDL by a factor of 2 to 10, serving as a
“safety factor,” the commercial laboratories hope to quantify
the environmental sample concentrations with a degree of
certainty. The degree of the factor (2-10) is decided by the
analytical laboratory depending upon the skill and experience
of the analyst, the quality of the instrument, and the nature
of the sample objectives. There is also another term called
reporting limit (RL), which is a limit imposed upon the
reporting laboratory. RL is usually demanded by the client or
regulatory guidelines, and basically associated with MDLs or
PQLs. If a sample
concentration is detected between MDL and PQL, the analytical
laboratory usually claims that this compound can be identified
but cannot be quantified in concentration in a laboratory
self-proclaimed confidence that already has a mathematically
calculated 99% confidence level (MDL) (see equation (1)). In
this case, the concentration between MDL and PQL is estimated
with the indicator “J”, which is called “J-value.” Figure 2
depicts the relative positions among MDL, PQL, EQL, and
J-value. However, although J-value may not provide the
quantification of a concentration, the result may still be
useful. The following real case example is a typical
environmental forensics case using J-value. The case is a
California groundwater contamination impacted by three
gasoline stations (A, B, and C) at an intersection, with a
co-mingled gasoline plume. There are three groundwater layers
underlying Station A: the perched water, shallow aquifer, and
deep aquifer. The co-mingled plume has been detected at the
deep aquifer. Groundwater data indicate that gasoline
constituents are detected relatively high at the perched
water, only J-values at the shallow aquifer, and also
relatively high at the deep aquifer. Therefore, Station A
easily argued that Station A only contributes contamination to
the perched water layer, not the co-mingled plume in the deep
aquifer since the shallow aquifer between the two water layers
is relatively non-impacted underlying its station. However,
further review of the data found that a low concentration of
gasoline oxygenate additive tertiary amyl methyl ether (TAME)
has been detected in J-values in the shallow aquifer.
Furthermore, TAME has been detected in the perched water and
the deep aquifer underlying Station A. This oxygenate compound
TAME is known as a unique gasoline additive by a certain brand
gasoline refinery, particularly marketed in California; and in
this case sold at Station A. Although the analytical result
cannot confidently quantify the concentration of TAME in the
shallow aquifer in this case, the J-value identification can
still lead investigation to the linkage of contaminant
downward movement from Station A to the perched, through the
shallow aquifer, and to the deep aquifer co-mingled plume.
Therefore, the J-value is used as the linkage piece in the
investigation.
Given the definitions of all
these detection limits, the laboratory reporting “non-detect”
really means an estimate concentration below the “detection
limit.” The detection limit can be MDL or PQL, and will be
determined by the end-users. Therefore, knowledge of detection
limit is necessary to review a laboratory report. For example,
a laboratory report indicates benzene concentration in
discharge effluent is non-detect with MDL = 0.11
mg/L
and RL = 2
mg/L.
Given the discharge limit is 1
mg/L
for benzene in the permit, a regulator determined that the RL
= 2
mg/L
indicated a non-compliance with the discharge limit. However,
in this case, laboratory footnotes specify that the non-detect
really means analyte not detected above MDL, which is 0.11
mg/L
and meets the 1
mg/L
limit. Therefore, the definition of “detection limit” is
crucial to review the data prior to making a decision.
People often ask why
regulations cannot make things easier by simply specifying a
“detection limit” for everyone to comply with in all cases. In
general, if we want detection limit too low, we may compromise
the analytical certainties in environmental sample
quantification; if we set forth detection limit too high, we
may not be able to detect the target contaminants in low
levels and also lose the usefulness of the J-values.
Therefore, regulatory requirements on laboratory detection
limits must be task specific. For example, when we are dealing
with waste water discharge permit to surface water bodies, the
detection limit must be low enough to meet the limits
permitted. When we are doing environmental site investigation,
any low concentration of contaminants may provide a clue in
terms of pollution distribution and migration; therefore,
detection limits should be low enough to generate useful data.
On the other hand, when we test hazardous waste, which we know
has very high concentrations in the first place, low detection
limit may not be necessary. Therefore, the requirement on
detection limit depends upon what are the target compounds and
sampling objectives. The goal of the necessary detection limit
should be to maximize the lowest level needed for the sample
objective. The knowledge of different type of detection limits
is essential for environmental forensics.
Disclaimer: The viewpoints expressed here are
merely author’s opinion and do not represent the position of
the regulatory organization that the author works for.
References
Code of Federal Regulations (CFR)
40 Part 136, Appendix B, pp554-555. (1993). U.S. Government,
Washington D.C. 20402-9328.
Pavia, D.L., Lampman, G.M., Kriz,
G.S., and Engel, R.G. (1995). Introduction To Organic
Laboratory Techniques (2nd Ed). Saunders
College Publishing, Harcourt Brace & Company, 8th
Floor, Orlando, FL 32887. |
