CAN LINDANE, ENDRIN, AND METHOXYCHLOR BE ISOLATED AND RECOVERED USING RP-SPE?

Sample Preparation Techniques



243



with an Saq of between 7.5 and 10 ppm. Endrin is a member of the cyclodiene

insecticides, possessing the characteristic endomethylene bridge structure. The Saq

for endrin is 0.23 ppm, which is similar to that of aldrin, dieldrin, and heptachlor

epoxide, but greater than that of heptachlor and chlordane. Endrin is known to

partially break down to endrin aldehyde and endrin ketone in packed GC columns.

It is not much of a problem when capillary columns are used. Methoxychlor, a

substituted diphenyl trichloroethane similar to DDT, has a Saq of between 0.1 and

0.25 ppm, and it is about 100 times more soluble in water than DDT. Using ethyl

acetate as an elution solvent and some of the early chemically bonded silica, of 60 Å

pore size and 40-µm irregular particle size (Separalyte, Analytichem International),

this author conducted a series of systematic studies of lindane, endrin, and methoxychlor (L, E, M). The work was focused on a sample matrix consisting of distilled

deionized water (DDI), and the analysis was performed on a gas chromatograph that

used a packed column and electron-capture detector.

One such study is reported here. Nine cartridges were packed with an octylbonded silica, C8 Separalyte. The mass of the sorbent packed in each cartridge varied

from a low of 20 mg to a high of 320 mg. In each cartridge, after methanol

conditioning, a spiked aqueous sample was passed through the packed bed under a

reduced-pressure SPE manifold. Then 8.8 µL of a methanolic reference standard

containing L, E, and M, each at a concentration of 100 ng/µL (ppm), was added to

approximately 60 mL of DDI. This spiking gave a concentration level in the aqueous

sample of 15 pg/µL (ppb). This concentration level is considered low enough to

approximate the realm of TEQA. After the sample was passed through the nine

replicate RP-SPE cartridges, two 500-µL aliquots of ethyl acetate were used to elute

the sorbed analytes into a 1-mL receiver volumetric flask. The contents of the

volumetric flask were quantitatively transferred to a 2-mL GC autosampler glass

vial, and 2 µL of this eluent was injected into a GC that contained a packed

chromatographic column. Only one injection per GC vial was made in this particular

study, so that variation in the percent recovery reflects only the random error associated with the RP-SPE process only. Table 3.12 lists the percent recovery for all

three OCs from each of the nine cartridges. The concentration of analyte that

constituted a 100% recovery was calculated and not actually measured; hence, there

is no contribution to the relative standard deviation from the control standard, as

shown in Equations (2.30) and (2.31). It is evident from review of the percent

recovery results shown in Table 3.12 that breakthrough was not reached even in the

case where only 20 mg of sorbent was taken to prepare the packed cartridge.

The mean percent recovery can be found from these nine replicate SPEs. The

standard deviation in the mean percent recovery is then found using the fundamental

equation for calculating standard deviations. If replicate injections per GC vial were

made, a pooled standard deviation relationship, as given in Equation (2.33), would be

most appropriate. From the standard deviation, a relative standard deviation or coefficient of variation is obtained, followed by calculation of the corresponding confidence

interval. For the percent recoveries shown in Table 3.12, the following statistical evaluation was obtained:



© 2006 by Taylor & Francis Group, LLC



244



Trace Environmental Quantitative Analysis, Second Edition



Mean

% Recovery



OC

Lindane

Endrin

Methoxychlor



RSD

(%)



Confidence Interval

(95%)



96.8

107.3

94.1



8.3

14.3

8.2



6.2

11.8

5.9



TABLE 3.12

Percent Recoveries of OCs Using a C8-Bonded

Silica Sorbent from Spiked Water

Sorbent (mg)



Lindane



Endrin



Methoxychlor



115

291

225

241

98

20

52

130

320



94.3

97.1

106

100

100

100

77.1

97.1

100



111

87.3

133

111

122

106

87.3

113

95.2



93.2

80.8

107

100

100

93.2

86.3

93.2

93.2



Note: 8.8 µL of 100 ppm each of lindane, endrin, and methoxychlor, dissolved in methanol, was added to approximately

60 mL of distilled deionized water. The spiked samples were

passed through the cartridge, previously conditioned with

methanol, and eluted with two 500-µL aliquots of ethyl acetate.

The volume of eluent was adjusted to 1.0 mL, and 2 µL of

eluent was injected into a GC.



The mean percent recoveries in this replicate series of RP-SPEs are very high

and represent a statement of accuracy in the measurement. A relative standard

deviation of 8 or 14% among replicate SPEs represents the precision of the method.

The confidence interval states that of the next 100 SPEs, 95 of these should fall

within the interval specified. For example, if 100 additional percent recoveries using

the SPE method could be performed for lindane, one could expect that 95 would

fall within 96.8 ± 6.2%.

This high and reproducible percent recovery for the isolation and recovery of

lindane from water strongly suggests that RP-SPE is very appropriate as a sample

preparation method for this analyte. Lindane is of continued interest to environmental

and toxicological scientists. One such study, discussed next, taken from the author’s

work, involves the isolation and recovery of lindane from homogenized myometrial

tissue suspended in an aqueous matrix.

© 2006 by Taylor & Francis Group, LLC



Sample Preparation Techniques



245



68. WAS LINDANE ISOLATED AND RECOVERED FROM A

BIOLOGICAL MATRIX USING RP-SPE? IF SO, HOW?

One hundred microliters of a methanolic solution containing 5 ng/µL (ppm) lindane

was placed in a 1-mL volumetric flask half filled with iso-octane, while 100 µL of

the same lindane reference standard was added to approximately 70 mL of distilled

deionized water (DDI). One microliter of the former solution containing 500 ng of

lindane was injected into a GC, and the resulting peak area served to define a control

that represents a 100% recovery of lindane. One microliter of the 1-mL eluent from

performing the RP-SPE of the spiked DDI was also injected into the same instrument,

and the resulting peak area served to define the spiked recovery sample. In this study,

quadruplicate injections of the control yielded a mean concentration of 400 ppb

lindane from interpolation of the least squares calibration curve. Seven replicates of

the eluent from the spiked DDI sample were injected, and a mean concentration of

406 ppb lindane was obtained from interpolation of the same calibration curve.

Equation (2.57) was used to find a 106% recovery of lindane. RSDs in the % recovery

were found by propagating random error between spiked samples and control reference standards. A complete result can be given by stating both the accuracy, 106%

recovery, and the relative standard deviation of 15.6%. Myometrium samples were

then prepared and lindane appeared as expected. This work showed that lindane

could easily be isolated and recovered from an aqueous matrix and confirmed the

earlier work on lindane isolation and recovery, discussed previously. One would be

led to believe that this is a robust sample preparation method for lindane because it

can be reproduced with confidence. Thus, a subsequent request for additional sample

analyses does not require extensive QC and should merely report the analytical

results.



69. WHAT DOES A SAMPLE ANALYSIS REPORT LOOK LIKE?

In addition to a tabular format for the determination of lindane in each of the

myometrial tissue samples submitted, a method summary should be included. A

one-paragraph method summary serves to inform the reader as to how the samples

were handled once they arrived in this analyst’s laboratory. The summary should

also provide a brief overview of the sample preparation and a brief description of

the determination technique used. The following report illustrates these concepts.

REPORT ON THE QUANTITATIVE DETERMINATION

TISSUE SUSPENDED CELLS IN SALINE



OF



LINDANE



IN



MYOMETRIAL



Summary of Method

The entire contents (1 mL) of sample that was received by the client were refrigerated

upon receipt until the sample was prepared for analysis. Upon thawing, the entire

contents of each sample were added to a reservoir that contained approximately 70 mL

of distilled deionized water. This aqueous solution was passed across a previously

conditioned octadecyl-bonded silica sorbent (C18RPSiO2). The retained analyte was

© 2006 by Taylor & Francis Group, LLC



246



Trace Environmental Quantitative Analysis, Second Edition



eluted off of the sorbent with two 500-µL aliquots of pesticides-residue-grade isooctane. The iso-octane eluent was then passed through a second SPE cartridge. This

second cartridge was packed with approximately 0.5 g of anhydrous sodium sulfate.

The volume of eluent that now contained the recovered analyte was adjusted to a final

volume of 1.0 mL using a volumetric flask. Three microliters of this eluent was injected

via autosampler into an Autosystem Gas Chromatograph (PerkinElmer) incorporating

an electron-capture detector. A 30 m × 0.32 mm capillary GC column containing DB-5

(J&W Scientific) was used to separate the organics in the eluent. This instrument is

abbreviated C-GC-ECD to distinguish it from other gas chromatographs in our laboratory. The column was temperature programmed following injection from 200 to

270°C at a rate of 10°C/min. The C-GC-ECD is connected to a 600 Link (PE-Nelson)

interface module. This interface, in turn, is connected to a 386 personal computer. This

PC uses Turbochrom® (PE-Nelson) software for data acquisition, processing, and

control. A method specific for lindane was written, and one peak was identified within

a 3% relative time interval (retention time window). A retention time, t[R], of 1.8 min

was consistently reproduced using autosampler injection.



Calibration

A series of calibration or working standards were prepared from the methanolic stock

solution containing lindane. This stock solution was prepared by carefully weighing

out pure solid lindane on an analytical balance. These standards were injected into the

C-GC-ECD from lowest to highest lindane concentration. The peak at 1.8 min was

identified as a reference peak in the Turbochrom software, and a narrow t[R] window

was defined around the t[R] of the apex of the peak. A calibration curve was constructed

using a least squares regression algorithm in the software. A correlation coefficient of

0.9990 was obtained.



Sample Analysis

The sequence within which the samples were run after calibration was created by using

the Sequence File Editor within Turbochrom. Samples are injected via autosampler

according to the instructions in the Sequence File. After the sequence is completed, a

Summary File is created and the analytical results are reported within this summary

format. In the Summary File, for each sample analyzed, the following is given:

(a) The sample number

(b) The concentration of lindane in ppb for 1.0 mL of eluent, interpolated from the

external standard mode of instrument calibration

(c) The retention time for lindane, t[R], in minutes

(d) The integrated peak area, in microvolts-seconds

The reported concentration is that for each eluent and should be multiplied by the

eluent volume (in this case, 1.0 mL) to obtain the number of nanograms of lindane

found. The number of nanograms found divided by the volume of the aqueous sample

used in RP-SPE gives the reported concentration in ppb for lindane. The reported

concentration should be divided by the volume of the sample to obtain the reported

concentration in ppb for lindane. The reported concentration should also be divided

by the percent recovery, expressed as a decimal, in order to find the true and final

concentration of lindane in the original sample.

© 2006 by Taylor & Francis Group, LLC