AGRICULTURAL MATERIALS

Determination of Selenium in Feeds and Premixes: Collaborative

Study
PALMER & THIEX: JOURNAL OF AOAC INTERNATIONAL VOL. 80, NO. 3, 1997
IVAN S. PALMER AND NANCY THIEX
South Dakota State University, Department of Chemistry and Biochemistry, Brookings, SD 57007
Collaborators: R. Allen; E. Alley; N. Anderson; J. Bell; N. Carpenter; W. Cunningham; L. Deuschle; P. Kane; R. Marts; G.
Rottinghaus; S. Rutta; L. Torma; A. Vindiola; J. Wenger; P. Whanger; A. Williams; C. Wo

A total of 17 laboratories participated in a collaborative study for the determination of selenium in

feeds and premixes using either a fluorometric or
a continuous hydride generation atomic absorption (HGAA) method. Each collaborator analyzed
16 blind duplicate samples of feed and premixes
from various feed manufacturers. The amount of
Se in these materials ranged from 0.2 to 5500
g/g. Six laboratories used only the fluorometric
procedure, 8 laboratories used only the hydride
generation atomic absorption procedure, and
3 laboratories used both procedures. One laboratory in the fluorometric study and 3 laboratories
in the HGAA study were initially excluded because of invalid data. Poor agreement between
the blind duplicates indicated probable sample interchange and/or dilution error. The data from
8 laboratories were submitted to statistical
analysis, including data from 2 laboratories participating in both studies. The repeatability
standard deviation (RSDr) for samples analyzed
by the fluorometric procedure ranged from 5.9
to 33%
%, and the reproducibility standard devia%. RSDr for
tion (RSDR) ranged from 12 to 33%
samples analyzed by HGAA ranged from 2.8 to
%. Both
18%
%, and RSDR ranged from 4.0 to 36%
fluorometric and continuous hydride generation atomic absorption methods for the determination of Se in feeds and premixes have been
adopted first action by AOAC INTERNATIONAL.

o maximize animal performance, it is important to monitor

the level in feeds of micronutrients such as selenium. It is
necessary to have analytical methods that are adequate for

Submitted for publication July 15, 1996.

The recommendation was approved by the Methods Committee on
Feeds, Fertilizers, and Related Agricultural Topics, and was adopted by the
Official Methods Board of the Association. See Official Methods Board
Actions (1996) J. AOAC Int. 79, 111A, and Official Methods Board
Actions (1996) The Referee, October issue.

measurement of the element in finished feeds. It is also necessary that the methods be suitable for the measurement of the
element in the premixes used to adjust diets to optimum Se
levels.
Several methods have been developed for Se analysis, including neutron activation analysis, X-ray fluorescence,
fluorometry, various forms of atomic absorption, and inductively coupled plasma spectrometry (1, 2). Some methods require instrumentation not available in every laboratory, and
others have limited detection limits.
The purpose of this study was to quantitate levels of Se in
feeds and premixes by using a fluorometric method or a continuous hydride generation atomic absorption (HGAA)
method. The sample oxidation procedure and fluorometric
method closely follow the procedure of Olson et al. (3) with
modifications suggested by Koh and Benson (4). The sample
oxidation procedure for the HGAA method is identical to that
used in the fluorometric method, and it is best described by Koh
and Benson (4). Instrumental operating conditions are described in the instrument manual (5). Both methods can be used
for determination of total Se in complete feeds and various premixes.
Collaborative Study
In a preliminary study, protocols and 4 samples, ranging in Se
concentration from 1 to 100 g/g, were sent to laboratories that had
indicated interest in participation. The purpose of this preliminary
study was to discover whether any problems existed in understanding the protocol. Laboratories were given 1 month to return
results. They were encouraged to identify and clarify difficulties in
the method prior to reporting results. After results were received
from the preliminary study, the expected values were shared with
the participating laboratories and difficulties and unclear aspects
of the procedures were discussed. The procedures were then corrected as necessary.
In the final study, the corrected proposed procedures and
16 blind pairs of samples (a total of 32 samples) were submitted to
17 laboratories for Se analysis. Samples included finished feeds
and premixes for cattle, swine, poultry and 2 finished pet foods.
Table 1 gives a short description of the samples. Samples were

grouped into the following 4 Se concentration ranges: 0.22

g/g, 210 g/g, 10100 g/g, and 1005500 g/g. To avoid
contamination and to aid in planning dilutions, laboratories
were informed of the concentration range for each sample.
Nine laboratories analyzed samples by the fluorometric
method, and 11 laboratories analyzed samples with the HGAA
method. Three laboratories used both methods. Laboratories
were instructed to perform a single analysis for all samples.

Preparation of Samples
At least 5 kg of each of 20 materials was obtained from feed
manufacturers. Each material was randomly numbered and
ground to pass a 1.0 mm screen in a Brinkman Retsch Mill
(Model AM1) equipped with a flow-through device. The resulting powder was mixed well and divided into 20 subsamples
and 20 blind duplicates by using a gated riffle.
To determine adequacy of mixing, 6 samples (3 subsamples
and 3 blind duplicates) from each feed or premix sample were
chosen randomly and analyzed. A separate coefficient of variation (CV) was calculated for each set of 3 values. For the
40 sets of samples and blind duplicates, CV was 6.2% or less
for all but 3 pairs. The remaining 3 pairs were analyzed a second time. In the second set of analyses, 2 pairs showed a CV
<5%. These were considered adequately mixed for inclusion on
the study. The material represented by the third pair was discarded because of nonuniformity. The mean CV for the 38 subsamples and blind duplicates available for the study was 2.15.
From the 19 materials passing the uniformity test, 16 were
chosen for the collaborative study. These included 5 samples in
the 0.22.0 g/g range, 4 samples in the 2.010 g/g range, 3 samples in the 10100 g/g range, and 4 samples in the 1005500
g/g range. Sixteen samples and their blind duplicates (32 samples) were submitted to each collaborator. In addition to samples,
each participating laboratory was given a commercially available
atomic absorption standard containing a known amount of Se (approximately 1000 g/mL). The standard was validated by comparison to Standard Reference Materials (SRM) 3149 and SRM 3172a.
996.16 Selenium in Feeds and Premixes
First Action 1996
(Applicable to determination of total Se in range of 0.2
5500 g/g or greater in all types of feed ingredients, finished
feeds, and premixes. Previous studies have shown it to be
adaptable to levels as low as 0.02 g/g. Estimated detection
limit is 4 ng Se/g.)
Caution: See Appendix B, safety notes on safe handling of
acids- nitric and perchloric acids, and safe handling of special
chemical hazards- cyclohexane. 2,3-Diaminonaphthalene is
highly toxic and carcinogenic. It causes eye and skin irritation,
is irritating to mucous membranes and upper respiratory tract,
and may be fatal if swallowed. When handling powders, wear
gloves and protective clothing and appropriate NIOSH/MSHA
approved respirator and perform operations in a fume hood.
Wash thoroughly after handling. Keep 2,3-diaminonaphthalene
tightly closed in cool dry storage area. In case of contact, im-

mediately flush eyes or skin with copious amounts of water and

discard contaminated clothing. If inhaled, remove to fresh air.
If not breathing, give artificial respiration. If breathing is difficult, give oxygen. If swallowed, wash mouth with water provided person is conscious, and call physician. Sodium borohydride is a flammable, toxic solid that reacts with water or acid
to release flammable hydrogen gas. Avoid contact with skin
and store in cool dry place. Destroy reagent before disposing.
Dispose of waste solvents in an appropriate manner compatible
with applicable environmental rules and regulations.
Alternative I: Fluorometric Method
Method Performance:
See Tables 996.16A and B for method performance data.
Accuracy of method was substantiated by in-house analyses of
NIST Standard Reference Materials (SRMs). Results of analyzing 5 replicates of 3 SRMs were as follows (numbers in parentheses are the certified values):
NIST-1567, wheat flour: 0.964 0.011 (1.1 0.2)
NIST-1577a, bovine liver: 0.692 0.020 (0.71 0.07)
NIST-1643c, trace elements in water: 0.0120 0.0002
(0.0127 0.0007)

A. Principle
All forms of Se in test sample are converted by digestion/oxidation procedure to inorganic form, presumably Se4+ or
Se6+. Presence of perchloric acid in oxidation mixture prevents
loss of Se. Se must be in Se4+ valence to react with 2,3-diaminonaphthalene. All forms of Se in digestion mixture are converted to Se4+ by reduction with HCl. Digestionreduction procedure is also suitable for quantitation of Se by continuous
hydride generation atomic absorption (HGAA) method (alternative II). Samples containing high amounts of soluble iron
(1 mg Fe/aliquot analyzed) may cause difficulty with
fluorometric procedure, but problem usually can be visually
detected and corrected by dilution or by increasing amount of
chelating agent.

B. Apparatus
(a) Fluorometer.Capable of excitation at 375 nm and
emission at 525 nm. If possible, adjust fluorometer so that
1 scale unit = 1 ng.
(b) Fume hood.Suitable for handling perchloric acid.
(c) Digestion system.21 26 7.4 cm aluminum block
with 80 holes (22 mm diameter) set on 30 30 cm hot plate.
(Any commercially available heating block is suitable, if test
samples and Se standard solutions can be heated simultaneously.) Alternatively, micro-Kjeldahl digestion system capable
of holding 30 mL flasks or straight-walled tubes may be used.
(d) Digestion vessels.Suitable for digestion system;
screw capped (Teflon lined) 20 150 mm test tubes. MicroKjeldahl flasks, 30 mL, or straight-walled tubes are acceptable.
(e) Extractor.Mechanized rotation unit, capable of maintaining 6070 rpm. Hand-held container allowing mixing of
rack (4 rows of 10 tubes) of tubes is suitable.
(f) Pipettor.Capable of delivering 5 mL 1%.

(g) H2O baths.(1) Capable of maintaining 60 2C, and

(2) capable of boiling.
(h) Vortex mixer.
(i) Volumetric flasks.100 and 1000 mL.
(j) Erlenmeyer flasks.2501000 mL and 2 L.
(k) Filter paper.Qualitative paper, 11 m retention.

C. Reagents
Reagents should be at least analytical grade. Use deionized
H2O distilled in glass for preparation of solutions and dilutions.
(a) Cyclohexane.
(b) Hydrochloric acid (HCl) solution.0.1M. Pipet 8.3 mL
concentrated HCl into 1 L volumetric flask and dilute to volume
with H2O. Proportionate amounts for any convenient volume may
be used.
(c) Nitric acid (HNO3).70%.
(d) Perchloric acid (HClO4).70%.
(e) 2,3-Diaminonaphthalene (DAN) reagent.Weigh 1.0 g
DAN powder (97% purity) and transfer to 2 L Erlenmeyer flask.
Add 500 mL 0.1M HCl and warm 15 min in water bath set at
60C. Stir solution to help dissolve powder. Dilute to 1 L with
0.1M HCl. Extract solution 35 min with 4050 mL cyclohexane
and then discard cyclohexane layer. Repeat extraction 3. Filter
DAN reagent through filter paper pre-wet with 0.1M HCl. DAN
reagent is stable at least 2 weeks, when protected from light and
stored in refrigerator.
(f) (Ethylenedinitrilo)tetraacetic acid (EDTA) solutions.
(1) EDTA stock solution.0.1M. Place 37.2 g (ethylenedinitrilo)tetraacetic acid, disodium salt, into 1 L volumetric flask and dilute
to volume with H2O. (2) EDTA working solution.0.01M. Depending on number of tubes to be analyzed, dilute sufficient volume EDTA standard stock solution 10-fold with H2O to provide
15 mL/tube (e.g., 600 mL for 40 tubes).
(g) Selenite standard solutions.(1) Selenite standard stock
solution.0.4 g Se/mL. Pipet 1.00 mL selenite standard solution
(1000 g Se/mL in 1% HNO3; commercially available atomic absorption standard solution) into 1 L volumetric flask and dilute to
volume with 0.1M HCl. (Note: As alternative, dissolve appropriate
amount of elemental Se in HNO3 and then dilute with 0.1M HCl
to desired concentration.) From this solution, pipet 40 mL into
100 mL volumetric flask and dilute to volume with 0.1M HCl. (2)
Selenite calibrating standard solutions.Pipet 0.00 (reagent
blank), 0.200, 0.500, 1.00, and 1.50 mL selenite standard stock
solution, into separate digestion vessels, to obtain 0.00, 0.08,
0.200, 0.400, and 0.600 g Se/vessel.
(h) Sodium selenate (Na2SeO4) standard solution.0.4 g
Se/mL. Transfer 0.1915 g anhydrous Na2SeO4 into 1 L volumetric
flask, dilute to volume with H2O. Mix well. From this solution,
pipet 5.00 mL into 1 L volumetric flask and dilute to volume with
H2O.

D. Quality Assurance
Starting with digestion procedure, with each set of test samples
run 2 reagent blanks and at least 4 selenite calibrating standard
solutions, C(g)(2)(e.g., 0.080, 0.200, 0.400, and 0.600 g Se/vessel), and 1 tube containing 0.500 mL sodium selenate solution,
C(h)(0.2 g Se/vessel), to check adequacy of reduction step, se-

lenate does not react with DAN. Recoveries of 95105% should

be expected. Otherwise, reanalyze the entire set.
Other appropriate SRMs can be included in analysis, e.g.,
NIST 1643c, trace elements in water (most convenient to use),
NIST 1567a, wheat flour; and NIST 1577b, bovine liver. To
control analysis cost, omit predigestion steps for SRMs. Transfer
or weigh appropriate amounts of SRMs directly into digestion
tubes.

E. Determination
(a) Predigestion.Weigh ca 10 g premix or feed sample into
2501000 mL Erlenmeyer flask and record weight to the nearest
10 mg (Ws). (The largest flask feasible should be used to minimize
foaming problems.) Add slowly and carefully 75 mL HNO3 and
boiling chip (or several glass beads). (Caution: Samples with large
amounts of limestone or easily oxidizable materials may foam
when nitric acid is added.) Heat sample on hot plate until as
much of material as possible is in solution and nitric oxide
fumes subside (usually 15 min are adequate).
Cool solution and dilute quantitatively with H2O so that Se
content falls between 0.040.60 g/mL. Record final volume
of diluted predigest solution to the nearest mL (Vt).
(b) Digestion.If block digestion system is not available,
perform digestion and reduction procedures in micro-Kjeldahl
flasks using alternative digestion and reduction procedure, G.
Otherwise, proceed as follows:
(1) Thoroughly mix predigest solution from (a) to suspend
all undissolved materials. Pipet 1.00 mL aliquots into test tubes
(digestion vessels). If Se content of predigest solution is low
(<0.02 g/mL), aliquots up to 10 mL can be used. Record volume to nearest 0.01 mL (Va).
(2) Add a porous boiling bead to each tube, including
blanks, selenite calibrating standard solutions, and sodium selenate standard solution. If glass beads are used, add 23 beads.
(3) Add 4 mL HNO3 and 1 mL perchloric acid [5 mL
HClO4HNO3 (1 + 4, v/v)] to each tube.
(4) Place tubes in aluminum heating block. Raise temperature slowly to 210C (ca 2 h). White fumes of perchloric acid
should be visible in tubes at completion of digestion. After
reaching white-fume state, heat additional 15 min.
(5) Remove tubes from heating block. Cool tubes to room
temperature and heating block to 110150C.
(c) Reduction.Add 0.5 mL concentrated HCl to tubes
from (b)(5).
Place tubes again in heating block and heat 30 min. Ensure
that temperature is maintained between 110 and 150C for the
entire period.
(d) Derivatization and quantitation.(1) Remove tubes
from heating block and allow tubes to cool. Tubes must be
maintained at room temperature. (Note: Procedure may be interrupted at any time, up to and including this step.)
(2) Add 15 mL EDTA working standard solution, C(f)(2),
and 2 mL DAN reagent, C(e), to test tube. (Note: The two solutions may be added simultaneously; however, they should not
be mixed together more than 10 min immediately before use or
precipitate will form.) Mix each tube well on vortex mixer, taking vortex to the bottom of tube at least 2.

(3) Place rack of tubes in H2O bath set at 60C and maintain
30 min. Ensure that H2O level is above level of reaction mixture.
(4) Remove rack from H2O bath and cool tubes 5 min in
running tap H2O.
(5) Add 5 mL cyclohexane to each tube. Cap tubes with
Teflon lined cap. Extract 510 min in rotating extraction unit
(6070 rpm/min). (Note: Extraction can be performed manually by shaking [inverting] rack of tubes for period of time that
gives maximum extraction.)
(6) Transfer cyclohexane layer into fluorometer cuvettes.
Ensure that solution is free of any suspended H2O droplets that
might adhere to wall of cuvette in light path.
(7) Set excitation wavelength of fluorometer at 375 nm and
emission at 525 nm. Zero fluorometer with cyclohexane and
read blank to judge quality of DAN reagent. If reading is
greater than 23 fluorescence units, DAN reagent should be
extracted again with cylcohexane. Zero fluorometer against
blank.
(8) Determine fluorescence (F) of selenite calibrating
standard solutions and calculate regression equation for standard curve. Use slope (k) in calculating Se concentrations in test
samples. Depending on equipment available this may be automatically done by built-in calibration procedure. [Note: Fluorescence response is linear when using selenite calibrating
standard solutions at concentrations described in C(g)(2).
Standards containing as high as 2 g Se/vessel have been used
and maintained linear relationship.]
(9) Determine fluorescence of test sample.

F. Calculations
Depending on support software of fluorometer used, calibration data, dilution factors, and sample weights may be stored
in computer, and then final content of Se [g/g (ppm)] may be
printed out. Report g Se/g to 3 significant digits.
When using manual system, calculate Se content in test
sample as follows:
Se, g/g =

1
F Vt

k Va Ws

where F = fluorescence reading; k = slope of regression line

obtained from plotting g Se (x-axis) vs fluorescence reading
(y-axis); Ws = weight of test sample, g; Vt = total volume of
diluted predigest solution, mL; Va = aliquot of predigest solution used in analysis, mL.

G. Alternative Digestion and Reduction Procedure

When block digestion system is not available, perform digestion and reduction procedures in micro-Kjeldahl flasks as
follows:
(1) Pipet 1.00 mL predigest solution from E(a) into 30 mL
micro-Kjeldahl flask. If Se content of predigest solution is low
(<0.02 g/mL), aliquots up to 10 mL can be used. Record volume to nearest 0.01 mL (Va).
(2) Add boiling bead to each flask. Add 4 mL HNO3 and
1 mL perchloric acid into each flask [5 mL HClO4HNO3 (1 +

4, v/v) can be used], including blanks, selenite calibrating

standard solutions, and sodium selenate standard solution.
(3) Place flasks on micro-Kjeldahl digestion rack with
fume collecting manifold connected to aspiration system that
collects most of the fumes from flasks. (Note: Manifold and
aspiration system may be omitted if perchloric acid hood is
available.) Adjust heaters to gently boil contents of flasks.
(4) Heat flasks until HNO3 has evaporated and white fumes
of perchloric acid are clearly visible in neck of flask. Heat flasks
15 min after white fumes of perchloric acid are observed.
(5) Remove flasks from digestion rack, cool to room temperature, and add 0.5 mL concentrated HCl.
(6) Place flasks in boiling H2O bath and maintain 30 min.
Remove flasks from H2O bath and cool to room temperature.
(7) Quantitatively transfer digest to test tubes fitted with
Teflon lined caps with 15 mL EDTA working solution, C(f)(2).
Add 2 mL DAN reagent, C(e).
(8) Proceed to E(d), Derivatization and quantitation.
Alternative II: Continuous Hydride Generation
Atomic Absorption (HGAA) Method
Method Performance:
See Tables 996.16C and D for method performance data.
Accuracy of method was substantiated by in-house analyses of
NIST Standard Reference Materials (SRMs). Results of analyzing
5 replicates of 3 different SRMs were as follows (numbers in
parenthesis are the certified values):
NIST-1567, wheat flour: 0.955 0.027 (1.1 0.2)
NIST-1577a, bovine liver: 0.704 0.022 (0.71 0.07)
NIST-1643c, trace elements in water: 0.0121 0.0009 (0.0127
0.0007)

I. Principle
Wet digestion procedure used for preparation of test samples
is identical to one used in fluorometric method (Alternative I).
Digestion/oxidation procedure converts all forms of Se to inorganic form, presumably Se4+ or Se6+. Presence of perchloric
acid in oxidation mixture prevents loss of Se. Se has to be in
Se4+ valence to be converted to volatile hydride. All forms of Se
in digestion mixture are converted to Se4+ by reduction with HCl.

J. Apparatus
(a) Atomic absorption (AA) spectrometer.Double beam,
equipped with deuterium lamp background corrector, Se hollow cathode lamp, and continuous hydride generation accessory. Operating conditions: warmup time, 20 min; acetyleneair flame: acetylene, 12 psi; air, 60 psi; wavelength, 196 nm;
lamp current, 10 mA; slit width, 1 nm; equilibration time, 30 s;
integration time, 5 s; replicates, 3. Continuous hydride generation accessory: nitrogen gas purge, 50 psi; flow rate of sample,
8 mL/min; flow rate of 0.6% NaBH4 + 0.5% NaOH,
1.1 mL/min; flow rate of 10M HCl, 1.1 mL/min.
(b) Fume hood.See B(b).
(c) Digestion system.See B(c).
(d) Digestion vessels.See B(d). Calibrate test tubes to
20 mL.

(e) Vortex mixer.

(f) Dispensing pipet.To deliver 19.0 mL. Equivalent
pipet is suitable.
(g) Volumetric flasks.100 and 1000 mL.
(h) Erlenmeyer flask.2501000 mL.

K. Reagents
Reagents should be at least analytical grade. Deionized
water distilled in glass should be used for preparation of solutions and dilutions.
(a) Hydrochloric acid (HCl) solutions.(1) 1 + 1, v/v.
For sample dilution. Dilute concentrated HCl with H2O and
place in dispensing pipet, J(f). (2)10 M. For hydride generation. Transfer 400 mL concentrated HCl into 500 mL volumetric flask and dilute to volume with H2O.
(b) Nitric acid (HNO3).70%.
(c) Perchloric acid (HClO4).70%.
(d) Sodium hydroxide (NaOH).
(e) Sodium borohydride solution (NaBH4).For hydride
generation. Dissolve 2.5 g NaOH and 3.0 g NaBH4 in 500 mL
H2O. (Caution: Sodium borohydride is a flammable, toxic solid
that reacts with water or acid to release flammable hydrogen
gas. Avoid contact with skin and store in cool dry place. Destroy reagent before disposing.)
(f) Selenite standard solutions.(1) Selenite standard
stock solution.1 g/mL. Pipet 1.00 mL Se standard solution
(1000 g Se/mL in 1% HNO3; commercially available AA
standard solution) into 1 L volumetric flask and dilute to volume with 0.1M HCl. (Note: As alternative, dissolve appropriate
amount of elemental Se in HNO3 and then dilute with 0.1M
HCl to desired concentration.) (2) Selenite calibrating standard
solutions.Pipet 0.500 mL, 1.00 mL, and 2.00 mL selenite
standard stock solution into separate 100 mL volumetric flasks
and dilute to volume with HCL solution (1 + 1, v/v) to obtain
Se concentrations of 5.00, 10.0, and 20.0 ng/mL.
(g) Sodium selenate (Na2SeO4) standard solution.0.4 g
Se/mL. See C(h).
(h) Quality control (QC) standard solution.8.00 ng/mL.
Pipet 40 mL selenite standard stock solution, K(f)(1), into
100 mL volumetric flask and dilute to volume with 0.1M HCl.
From this solution, pipet 2.00 mL into 100 mL volumetric flask
and dilute to volume with HCl solution (1 + 1, v/v).

M. Determination
(a) Predigestion.See E(a).
(b) Digestion.See E(b). Note: If block digestion is not
available, perform digestion and reduction procedures in micro-Kjeldahl flasks according to alternative digestion and reduction procedure, O.
(c) Reduction.After tubes from M(b) have cooled to
room temperature, add 0.5 mL concentrated HCl.
Place tubes again in heating block and heat 30 min. Ensure
that temperature is maintained between 110150C for the entire period.
Remove tubes from heating block and allow to cool to room
temperature. Dilute contents of tubes to 10 mL with HCl solution (1 + 1, v/v), mix well with Vortex mixer, or cap and mix by
inverting several times.
(d) HGAA determination.(1) Calibrate instrument with
HCl solution (1 + 1, v/v) as reagent blank and selenite calibrating standard solutions, K(f)(2). At beginning of analysis, calibration procedure should be repeated until absorbance values
(particularly for 5 ng/mL Se working standard solution) remain
constant. AA spectrometer response is not necessarily linear for
range of Se standard working solutions used in analysis. Concentration of unknowns must be determined from standard curve.
(2) Read QC standard solution as sample. Concentration of
8.0 0.4 ng Se/mL should be obtained. Otherwise, recalibrate
instrument.
(3) Determine content of Se (C, ng/mL) in test sample.
(4) Check Se content in QC standard solution every 10
12 samples. If Se content falls outside acceptable range (i.e.,
8.0 0.4 ng/mL), reslope calibration curve or recalibrate
instrument.
(5) If time allows, read each sample twice but not consecutively. Average 2 values. Readings should agree within 10%.
Greater deviation indicates mixing or standardization problem
and samples should be reanalyzed.
If sample reading is outside calibration curve, dilute portion
of test sample with HCl solution (1 + 1, v/v) and repeat analysis.
Use appropriate dilution factor in calculations.

N. Calculations
It is assumed that all samples are diluted to 20 mL (final
dilution) and that AA spectrometer is calibrated in ng Se/mL.
Calculate Se content, g/mL, as follows:

L. Quality Assurance
Starting with digestion procedure, with each set of test samples run 2 reagent blanks, and 1 tube containing 0.500 mL sodium selenate solution, K(g) (0.2 g Se/vessel), to check adequacy of reduction step, since selenate is not reduced to Se
hydride by NaBH4. Recoveries of 95105% should be expected. Otherwise, reanalyze the entire set.
Other appropriate SRMs can be included in analysis, e.g.,
NIST 1643c, trace elements in water (most convenient to use).
NIST 1567a, wheat flour; and NIST 1577b, bovine liver. To
control analysis cost, omit predigestion steps for SRMs. Transfer or weigh appropriate amounts of SRMs directly into digestion tubes.

Se, g/mL =

Vt
1
C
20

Va Ws
1000

where C = Se content in each tube determined by HGAA,

ng/mL; Vt = total volume of diluted predigest solution, mL; Va
= aliquot of predigest solution analyzed, mL; Ws = weight of
test sample, g.

O. Alternative Digestion and Reduction Procedure

If block digestion system is not available, perform digestion
and reduction procedures in micro-Kjeldahl flasks as follows:
(1) Pipet 1.00 mL predigest solution from M(a) into 30 mL
micro-Kjeldahl flask. If Se content of predigest solution is low

(<0.02 g/mL), aliquots up to 10 mL can be used. Record volume to nearest 0.01 mL (Va).
(2) Perform steps G(2)(6).
(3) After flasks have cooled to room temperature, dilute digests to 20 mL with HCl solution (1 + 1, v/v) and mix thoroughly. Transfer diluted digest to test tubes.
(4) Proceed to M(d), HGAA determination.
Ref.: J. AOAC Int. 80, 469480(1997)
CAS -771-97-1 (2,3-diaminonaphthalene)

The variation between laboratories (reproducibility relative

standard deviation; RSDR) was higher than anticipated on the basis
of the methods performance in the Associate Referees laboratory.
Eight of 16 sample pairs had RSDR values 25% (Table 996.16A).
Because the results of only 8 laboratories were submitted to the
statistical analysis for outliers, only one laboratory could be omitted
and still comply with the 2/9 rule. In several instances, the Cochran
and Grubbs tests (1% level) gave values very close to the significance level for excluding additional laboratories.

HGAA Method
Results and Disscussion
The commercial atomic absorption Se standard, C(g) in alternative I and K(f) in alternative II, was not traceable to NIST
standards according to the manufacturer. The Associate Referees laboratory compared the commercial standard with both
NIST SRM 3149 and SRM 3172a. The Se content of the commercial standard was found to be 101.5% of the expected concentration with NIST 3149 as the standard and 101.5% of the
expected concentration with NIST 3172a as standard.

Fluorometric Method
Results from 8 laboratories are presented in Tables 25. The
results from another laboratory were omitted before statistical
analysis because the poor agreement between a large number
of blind duplicates indicated probable sample interchange
and/or dilution errors and hence invalid data. From the remaining 8 laboratories, outliers were identified and eliminated by
using Cochran and Grubbs tests (1% level). Laboratories 16
and 26 had 3 and 2 pairs of results eliminated, respectively (Table 6), and laboratories 6 and 23 each had one pair of results
eliminated.
Deviations from the protocol were minor. Laboratory 28 did
not use the predigestion procedure, using sample sizes of 2
10 g instead of the prescribed 10 g size. Laboratory 26 used an
automated procedure for the steps after sample digestion and
reduction. Laboratory 16 reduced the sample size, using 5 g in
some instances. Laboratory 16 reported no data for sample 9
and 21.
Repeatability relataive standard deviation (RSDr) values
were more scattered than desired, varying between 5.9 and
33% as shown in Tables 996.16A and B. Samples in the lowest
Se concentration range (<2.00 g/g) showed the greatest variation, with 2 of 5 results having RSDr values greater than 25%.
RSDr values for only 2 other results exceeded 25%. These occurred in groups whose mean Se contents were 29 and 170
g/g. The preliminary uniformity study found the CV for the
samples ranged from 0.37 to 6.18% with a mean of 2.15%. A
small portion of the variation reflected in the repeatability value
can be accounted for by sample heterogeneity.
Several blind duplicates showed very large variations, for
example, results of laboratories 6 and 16 for samples 37, 39,
and 23, 8, respectively, (Table 2) and of laboratory 26 for samples 34, 5 (Table 3) and for samples 11, 28 (Table 4). Some of
these differences appear to be due to errors in labeling and/or
dilution factors used in calculating.

Results from 8 participating laboratories are shown in Tables 710. Results from 3 other laboratories were declared invalid
and omitted from the study before statistical analysis because
they each had several blind duplicates that did not agree, and
one laboratory did not have adequate sensitivity for samples 3740. From the remaining 8 laboratories, outliers were
identified and eliminated by using Cochran and Grubbs tests
(1% level). Laboratory 5 had 4 pairs of results eliminated, and
laboratories 15 and 28 each had one pair of results eliminated
(see Table 6).
Deviations from the suggested HGAA protocol were minor.
Laboratory 27 reported results on the basis of mg Se/100 g.
These values were converted to g Se/g. Laboratory 5 used
manual vapor generation to introduce sample in the AA system.
This modification was suggested in the comment section of the
original method protocol, but the one laboratory using this
modification had a large number of outlier values.
Performance parameters for results obtained by HGAA are
shown in Table 996.16C. RSDr values were <14% in all cases
except one. The highest variation tended to be in samples with
lowest Se concentrations, but the range of RSDr values was still
a very acceptable 4.418% (see Table 996.16D). The agreement between laboratories (reproducibility) showed greater but
still acceptable variation with a range of 4.036%. All but
4 RSDR values were under 25%, and the greatest variation was
obtained in samples with lowest Se concentrations.
The variability between laboratories would have been lower
if laboratory 5 had been excluded. Four of 6 outlier values
came from this laboratory. Several other results from this laboratory were low relative to the overall average, but they were
not considered as outliers at the 1% level of significance. Some
consideration was given to excluding laboratory 5 from all the
statistical analyses, but this would have resulted in a violation
of the 2/9 rule in 2 of the sample sets.
Comments
Some laboratories thought the predigestion procedure was
cumbersome and the amount of reagents required was too costly.
Others thought the subsequent dilutions were problematic. The
purpose of the predigestion steps was to use a large enough sample
so that sampling error was minimized and to reduce the amount of
organic matter exposed to perchloric acid digestion. If more effort
and time are expended in sample splitting and grinding, then
smaller samples can be used. However, increasing the organic material exposed to the perchloric acid will increase the potential for

charring, and, therefore, the digestion process will need to be

watched much more closely.
Recommendation
On the basis of the results of this study, it is recommended that
both the fluorometric and continuous hydride atomic absorption methods for the determination of Se in feeds and premixes
be adopted first action.

G. Rottinghaus, University of Missouri, Veterinary Medical

Diagnostic Laboratory, Columbia, MO
S. Rutta, Wisconsin Animal Health Laboratory, Madison, WI
L. Torma, Montana Department of Agriculture, Bozeman, MT
A. Vindiola, Office of State Chemist, College Station, TX
J. Wenger, Lancaster Laboratories, Lancaster, PA
P. Whanger, Department of Agricultural Chemistry, Oregon
State University, Corvallis, OR
A. Williams, Woodson Tenant Laboratories, Memphis, TN
C. Wo, Nutrition International, Dayton, NJ

Acknowledgments
We thank the volunteer participants in the collaborative
study for their willingness to undertake the effort:
R. Allen, Hazelton, Madison, WI
E. Alley, Mississippi State Chemistry Laboratory, Mississippi State, MS
N. Anderson, South Dakota State University, Brookings, SD
J. Bell, Barrow Agee Laboratories, Memphis TN
N. Carpenter, Pennsylvania Department of Agriculture, Summerdale, PA
W. Cunningham, Moorman Manufacturing Company,
Quincy, IL
L. Dueschle, Cornell University, Ithaca, NY
P. Kane, Office of the Indiana State Chemist, West Lafayette, IN
R. Marts, U.S. Food and Drug Administration, Shawnee
Mission, KS

References
(1) Thomassen, Y., Lewis, S.A., & Veillon, C. (1994) Techniques
and Instrumentation in Analytical Chemistry Series, Vol. 15,
R.F. Herber (Ed.), Ch. 22: Selenium, Elsevier Science Publishers, Amsterdam, The Netherlands, pp. 489500
(2) Alaejos, M.S., & Romero, C.D. (1995) Chem. Rev. 95, 227257
(3) Olson, O.E., Palmer, I.S., & Cary, E.E. (1975) J. Assoc. Offic. Anal. Chem. 58, 117121
(4) Koh, T.S., & Benson, T.H. (1983) J. Assoc. Off. Anal. Chem.
66, 918926
(5) Rothery, A.E. (1984) VGA-76 Vapor Generation Accessory Operation Manual, Varian Publication No. 85-100677-00, p. 17
(6) Olson, O.E., Palmer, I.S., & Whitehead, E.I. (1973) in Methods
of Biochemical Analysis 21, D. Glick (Ed.), John Wiley & Sons,
New York, NY, pp. 3975

Table 996.16A. Method performance for determination of Se in feeds and premixes by fluorometric method a
No. of labs after
removal of outliersb

No. of outlier labs

Nc

Dry dog food

Moist cat food
Medicated swine feed
Medicated swine feed
Swine feed supplement

7
7
8
8
7

1
1
0
0
1

14
14
16
16
14

0.312
0.379
0.88
1.16
1.45

0.049
0.039
0.23
0.31
0.25

16
10
27
27
17

0.093
0.10
0.24
0.31
0.36

30
27
28
27
25

0.14
0.11
0.66
0.86
0.69

0.26
0.29
0.69
0.86
1.0

Swine concentrate
Calf pellets
Swine starter premix
Swine supplement

7
8
7
7

1
0
1
1

14
16
14
14

1.97
2.39
3.93
7.65

0.16
0.44
0.38
0.66

8.1
18
9.6
8.6

0.33
0.44
0.87
0.94

17
18
22
12

0.45
1.2
1.1
1.8

0.52
1.2
2.4
2.6

Beef mineral/vitamin premix

Vitamin E-selenium fortifier (all livestock)
Selenium Pak

8
8
7

0
0
1

15
16
14

8.1
4.5
4.7

28
7.0
5.9

Selenium premix, 0.2%

Selenium premix, calcium carbonate, 0.06%
Selenium premix, 0.176%
Selenium premix, 0.6%

7
7
8
8

1
1
0
0

13
14
16
16

Sample

a
b
c
d
e

Mean, g/g

29.0
63.8
80.1
170
571
3130
5290

sr

56
88
500
350

RSDr, %

33
15
16
6.4

sR

8.3
9.6
13
56
150
940
760

Different Se concentration ranges are separated by line. In ascending order they are 0.22.0 g/g, 2.010 g/g, 10100 g/g, and 1005500 g/g.
Outliers determined by Cochran and Grubbs tests (1% level).
Number of samples, including blind duplicates.
r = 2.8 sr.
R = 2.8 sR.

RSDR, %

rd

Re

29
15
16

23
13
13

23
27
36

33
26
30
13

160
250
1400
990

160
420
2600
2100

Table 996.16B. Summary of method performance for determination of Se in feeds and premixes by fluorometry
Se, g/g
0.22.0
2.010.0
10.0100
1005500

sr

RSDr, %

sR

RSDR, %

0.0390.31
0.160.66
4.58.1
5650

1027
8.118
5.928
6.433

0.0930.36
0.330.940
8.313
56940

2530
1222
1529
1333

Table 996.16C. Method performance for determination of Se in feeds and premixes by HGAA method a
No. of labs after
removal of outliersb

No. of outlier
labs

Nc

Dry dog food

Moist cat food
Medicated swine feed
Medicated swine feed
Swine feed supplement

8
8
8
8
8

0
0
0
0
0

16
16
16
16
16

0.249
0.360
0.699
0.960
1.15

0.031
0.033
0.091
0.12
0.14

13
9.3
13
12
12

0.051
0.086
0.18
0.25
0.41

21
24
25
26
36

0.088
0.094
0.25
0.34
0.39

0.14
0.24
0.49
0.70
1.15

Swine concentrate
Calf pellets
Swine starter premix
Swine supplement

7
7
7
8

1
1
1
0

14
14
14
16

1.80
2.55
4.20
8.0

0.32
0.24
0.18
1.1

18
9.4
4.4
13

0.32
0.26
0.74
1.5

18
10
18
19

0.91
0.67
0.51
3.0

0.91
0.73
2.1
4.2

Beef mineral/vitamin premix

Vitamin E-selenium fortifier, all livestock
Selenium Pak

7
8
7

1
0
1

14
16
14

1.3
6.1
2.2

4.0
10
2.8

1.3
8.1
9.9

4.0
13
13

Selenium premix, 0.2%

Selenium premix, calcium carbonate, 0.06%
Selenium premix, 0.176%
Selenium premix, 0.6%

8
8
7
8

0
0
1
0

16
16
14
16

Sample

a
b
c
d
e

Mean, g/g

31.3
60.2
76.4
180
605
3070
5450

sr

20
69
120
540

RSDr, %

11
11
3.8
10

sR

47
110
540
540

Different Se concentration ranges are separated by line. In ascending order they are 0.22.0 g/g, 2.010 g/g, 10100 g/g, and 1005500 g/g.
Number of samples, including blind duplicates.
Outliners determined by Cochran and Grubbs tests (1% level).
r = 2.8 sr.
R = 2.8 sR

RSDR, %

26
18
6.1
10

rd

3.5
17
6.0
58
190
330
1520

Re

3.5
23
28
130
300
520
1520

Table 996.16D. Summary of method performance for

determination of Se in feeds and premixes by HGAA
Se, g/g
0.22.0
2.010.0
10.0100
1006000

sr

RSDr, %

sR

RSDR, %

0.0310.14
0.181.1
1.36.1
20540

9.313
4.418
2.810
3.811

0.0510.41
0.261.5
1.310
47540

2136
1019
4.013
6.126

Table 1. Feeds and premixes used

Expected Se
range, g/g
0.22.0

2.010.0

10.0100

1005500

Sample description

Sample No.

Dog food, dry

Cat food, moist
Medicated swine feed
Medicated swine feed
Swine feed supplement
Swine concentrate
Calf pellets
Swine starter premix
Swine supplement
Beef mineral/vitamin premix
Vitamin ESe fortifier (all livestock)
Se Pak

37, 39
38, 40
34, 24
36, 10
23, 80
14, 32
17, 26
13, 22
34, 52
11, 28
27, 38
31,15

Se premix (0.2%)
Se premix, calcium carbonate
(0.06%)

21, 16

Se premix (0.176%)

19, 18

Se premix (0.6%)

17, 20

35, 98

Table 2. Collaborative study results (

g/g) for determination of Se in feeds and premixes by a fluorometric method:
approximate Se content range of 0.2002.00 g/g
Sample number a
Laboratory No.
11
13
16
16
19
23
26
28
a
b

37, 39
0.270
0.317
1.25 b
0.580
0.270
0.273
0.295
0.320

38, 40
0.290
0.272
0.455
0.420
0.283
0.235
0.227
0.320

0.350
0.272
0.534
<0.200
0.401
0.387
0.387
0.280

0.380
0.365
0.617
<0.200
0.395
0.353
0.353
0.230

4, 24
0.630
1.27
1.41
0.540
0.827
0.711
0.683
1.00

36, 1
0.680
1.04
0.760
1.16
0.850
0.731
0.809
1.00

Values are presented as blind duplicates for each laboratory.

Underlined values were outliers as determined by Cochran and Grubbs tests (1% level).

0.940
1.49
1.15
1.67
1.13
1.14
1.11
1.28

23, 8
1.09
1.23
1.56
0.630
1.18
0.968
0.745
1.18

1.15
1.68
1.57
2.27
1.39
1.19
1.30
1.60

1.28
2.54
1.41
0.730
1.38
1.16
1.27
1.33

Table 3. Collaborative study results (

g/g) for determination of Se in feeds and premixes by a fluorometric method:
approximate Se content range of 2.0010 g/g
Sample number a
Laboratory No.
11
13
16
16
19
23
26
28
a
b

14, 32
2.18
2.12
1.55
b
1.32b
2.01
1.85
1.96
2.54

7, 1
2.40
1.90
1.53
2.56
2.04
1.66
1.57
2.28

1.97
3.16
2.26
3.01
2.46
1.99
2.12
2.02

13, 22
3.09
2.54
2.60
1.90
2.55
2.19
2.38
2.07

2.54
4.14
4.23
4.09
4.42
4.00
3.08
4.24

Values are presented as blind duplicates for each laboratory.

Underlined values were outliers as determined by Cochran and Grubbs tests (1% level).

34, 5
1.74
4.02
4.01
5.20
4.30
3.81
1.81
4.30

7.80
6.70
7.15
8.18
8.24
7.30
9.19
8.30

7.34
6.73
6.95
6.62
8.00
7.65
0.528
10.1

Table 4. Collaborative study results (

g/g) for determination of Se in feeds and premixes by a fluorometric method:
approximate Se content range of 10100 g/g
Sample number a
Laboratory No.
11
13
16
16
19
23
26
28
a
b

11, 28
28.6
29.6
14.9
26.1
31.8
28.0
b
4.96 b
25.0

27, 3
11.7
47.7
30.0
34.4
31.1
31.1
34.2
25.4

80.8
47.7
64.1
63.7
66.4
62.0
77.5
49.8

31, 15
71.8
61.6
61.4
65.6
64.7
63.3
71.9
48.6

73.3
66.3
70.5
81.2
79.6
68.4
82.9
108 1

109 1
78.6
79.9
75.7
80.9
66.9
77.8
105 1

Values are presented as blind duplicates for each laboratory.

Underlined value was omitted by the Associate Referee even though there was not significance in the outlier test. Because of the great
variation in duplicates, the laboratory was asked to check calculations. No response was made and no additional analyses were requested.
The value was excluded due to an assumed calculation or dilution error.

Table 5. Collaborative study results (

g/g) for determination of Se in feeds and premixes by a fluorometric method:
Se content of 1006000 g/g
Sample number a
Laboratory No.
11
13
16
16
19
23
26
28
a
b
c

21, 16
203
228
65.9
b b

189
590
149
175

35, 9
196
173
250
116
184
162
181
160

538
679
559
b
74.9c
649
608
415
620

19, 18
671
630
634

676
584
129
607

Values are presented as blind duplicates for each laboratory.

= Values were not supplied by laboratory.
Values were outliers as determined by Cochran and Grubbs tests (1% level).

3422
3246
1742
4560
3060
3060
3242
1730

17, 20
4138
3370
3057
3750
3160
3160
4027
1070

6194
5187
4863
7550
5740
5740
4993
4550

5472
5603
5143
6560
5130
5130
5322
4920

Table 6. Outlier laboratories identified by Cochran and

Grubbs tests for the fluorometric and HGAA methods of
Se analysis
Method of analysisa
Samples
37, 39
38, 40
14, 24
36, 1
23, 8
14, 32
17, 26
13, 22
34, 5
11, 28
27, 3
31, 15
21, 16
35, 9
19, 18
17, 20
a

HGAA
b

15
15
15

15

15

28

Fluorometric
16

16
16

26
26

23
16

Numbers denote laboratories whose data were deleted from

statistical analysis (1% level).
, no outliers detected.

Table 7. Collaborative study results (

g/g) for determination of Se in feeds and premixes by an HGAA method:
approximate Se content range of 0.2002.00 g/g
Sample number a
Laboratory No.
15
17
11
14
15
19
27
28
a

37, 39
0.240
0.251
0.212
0.320
0.154
0.271
0.160
0.260

38, 40
0.300
0.266
0.250
0.320
0.254
0.276
0.180
0.270

0.280
0.408
0.321
0.420
0.433
0.389
0.370
0.260

4, 24
0.280
0.435
0.339
0.510
0.411
0.379
0.360
0.170

Values are presented as blind duplicates for each laboratory.

0.560
0.895
0.848
0.650
0.684
0.750
0.600
0.860

36, 1
0.250
0.830
0.865
0.670
0.514
0.776
0.570
0.840

0.340
1.12
1.10
0.770
0.882
1.04
0.940
1.20

23, 8
0.660
1.09
1.15
0.740
1.19
1.18
0.850
1.18

0.460
1.47
1.44
0.740
1.04
1.29
0.970
1.82

0.380
1.41
1.42
0.820
1.41
1.34
0.990
1.42

Table 8. Collaborative study results (

g/g) for determination of Se in feeds and premixes by an HGAA method:
approximate Se content range of 2.0010 g/g
Sample number a
Laboratory No.
15
17
11
14
15
19
27
28
a
b

14, 32
1.40
1.87
1.80
2.30
1.60
2.04
1.70
1.33

7, 26
1.40
1.88
1.78
1.60
3.75
2.00
1.70
2.32

2.60b
2.69
2.48
2.70
2.85
2.37
2.70
2.94

13, 22
0.600
2.43
2.44
2.30
2.16
2.30
2.40
2.99

3.40
4.49
4.34
2.70
5.02
4.35
4.00
4.31

Values are presented as blind duplicates for each laboratory.

Underlined values were outliers as determined by Cochran and Grubbs tests (1% level).

34, 5
1.80
4.08
4.50
2.70
5.09
4.21
4.30
4.73

5.40
8.77
8.55
6.00
6.06
8.09
7.70
10.9

8.60
8.47
8.04
6.40
8.66
8.00
7.80
10.2 1

Table 9. Collababorative study results (

g/g) for determination of Se in feeds and premixes by an HGAA method:
approximate Se content range of 10100 g/g
Sample number a
Laboratory No.
15
17
11
14
15
19
27
28
a
b

11, 28
b

20.0b
32.4
32.0
30.0
32.8
32.0
30.7
28.6

27, 3
28.0
31.4
31.4
33.0
30.7
31.5
30.5
31.3

64.0
62.3
64.1
60.0
62.1
61.7
65.0
50.0

Values are presented as blind duplicates for each laboratory.

Underlined values were outliers as determined by Cochran and Grubbs tests (1% level).

31, 15
44.0
64.6
63.0
56.0
73.5
63.0
65.6
43.9

28.0
75.1
78.6
68.0
64.2
75.0
71.5
96.8

94.0
76.0
78.2
68.0
71.8
76.1
73.6
96.5

Table 10. Collaborative study results (

g/g) for determination of Se in feeds and premixes by an HGAA method: Se
content of 1005500 g/g
Sample number a
Laboratory No.
15
17
11
14
15
19
27
28
a
b

21, 16
105
177
157
190
273
190
179
176

35, 9
84.0
180
176
200
233
193
180
241

450
616
624
540
744
627
639
775

19, 18
350
635
632
570
511
657
627
680

3000
3162
3090
3100
2910
3180
2732
b
1860b

Values are presented as blind duplicates for each laboratory.

Underlined values were outliers as determined by Cochran and Grubbs tests (1% level).

17, 20
2900
3380
3130
3200
3250
3200
2802
1900

5600
5440
5360
5400
5800
5650
5128
7050

4500
5459
5410
5500
4860
5570
5047
5430