318 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO.

2, 2002

AGRICULTURAL MATERIALS

Determination of Water (Moisture) and Dry Matter in Animal

Feed, Grain, and Forage (Plant Tissue) by Karl Fischer Titration:
Collaborative Study
NANCY J. THIEX and TERRI VAN EREM
South Dakota State University, Oscar E. Olson Biochemistry Laboratories, Box 2170, ASC 151, Brookings, SD 57007

Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023

Collaborators: D. Amundson; J. Boucher; A. Clark; D. Clark; P. Coco; D. Dow; P. Fahey; H. Hoffmann; D. Mertens;
H. Morris; T. Rittenhouse; M. Wohl

A Karl Fischer method for determining water (dry
matter) in animal feed and forages was collabor-
atively studied. Water was extracted from animal
feed or forage material into methanol–formamide
(1 + 1) directly in the Karl Fischer titration vessel
by high-speed homogenization. The water was ti-
trated at 50°C with one-component Karl Fischer re-
agent based on imidazole. Ten blind samples were
sent to 9 collaborators in the United States, Can-
ada, and Germany. The within-laboratory relative
standard deviation (repeatability) ranged from 1.14
to 6.99% for water or from 0.09 to 0.56% for dry
matter. Among-laboratory (including within-) rela-
tive standard deviation (reproducibility) ranged
from 5.35 to 10.73%, or from 0.44 to 0.77% for dry
matter. The authors recommend that the method
be adopted as Official First Action by AOAC IN-
TERNATIONAL. A comparable alternative extrac-
tion procedure using boiling methanol is also rec-
ommended for Official First Action.
iscrepancies in results obtained when various oven
applicable for determination of water in dry, ground animal
feed and forage from 1–15% water; it is not applicable for
mineral mix feed. To recommend this KF method for deter-
mining water (moisture) in animal feed and forage (plant tis-
sue) as an Official Method, a collaborative study was con-
ducted to evaluate its repeatability and reproducibility. The
agricultural analytical community urgently needs a primary
moisture method for feed and forage material.
Principle
Feeds and forages are heterogeneous materials that contain
substances of cellular structures that release water very slowly
and incompletely. To speed the extraction and reaction, cellular
structures are ruptured by high-speed homogenization, and the ti-
tration is performed at an elevated temperature (50°C) in metha-
nol–formamide (1 + 1). The formamide dissolves sugar and
helps to distribute protein. Under this combination of conditions,
extractions and titrations can be completed in a short time (2).
Extraction of water from the animal feed or forage material
into boiling methanol directly in the KF titration vessel with-
out homogenization provides comparable recoveries of water,

D methods are used to determine moisture and dry mat-

ter in animal feeds were observed with American As-
sociation of Feed Control Officials (AAFCO) Check Sample
and is provided as an alternative extraction procedure for rapid
extractions and titrations. Other components extracted from
the samples with water, e.g., oils, and carbohydrates, do not
interfere with the KF method, which is based on a specific
Results and with National Forage Testing (NFTA) Profi-
chemical reaction that consumes water. This method provides
ciency Samples. In response to the need for a primary,
for the dry ground test portion to be weighed and added di-
nonempirical method for determining water in feeds and for-
rectly to the titration cell. This is the preferred method for reli-
ages by the agricultural community, a Karl Fischer (KF)
able administration of sample and minimal contamination by
method was developed and validated. The method validation,
atmospheric moisture.
including repeatability, reproducibility, and comparability
The chemistry of the KF reactions has been documented by
studies, was previously described in detail (1). The method is
Eugen Scholz (3):

Submitted for publication November 2001. CH3OH + SO2 +RN = [RNH] SO3CH3
The recommendation was approved by the Methods Committee on
Feeds, Fertilizer, and Related Agricultural Topics. Method I was approved
for Official First Action at their meeting in Kansas City, MO, on where RN = imidazole base.
September 8, 2001 and Method II was approved for Official First Action on
February 6, 2002. See “Official Methods Program Actions,” (2001) Inside
Laboratory Management, November/December issue. H2O + I2 + [RNH]SO3CH3 + 2 RN
Corresponding author’s e-mail: nancy_thiex@sdstate.edu. = [RNH]SO4CH3 + 2[RNH]I
 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002 319
The sulfur dioxide reacts with alcohol to form an ester,
which is neutralized by the base. The anion of alkyl sulfurous
acid is the reactive component and is present in the KF re-
agent. The titration of water constitutes oxidation of
alkyl-sulfate by iodine. This reaction consumes water. It
therefore follows that 2 significant prerequisites must be ful-
filled to ensure a stoichiometric course of the KF reaction. The
first of these is the presence of a suitable alcohol to esterify the
sulfur dioxide completely. Methanol is the preferred choice,
because the course of the KF reaction is stoichiometric and
rapid in methanol. In Method I, methanol is mixed with
formamide, which dissolves sugar and helps to distribute pro-
tein. The formamide content should not exceed 50% to main-
tain reaction conditions.
The second prerequisite is that a suitable base is necessary
for the complete neutralization of the acids produced during
the reaction. The pyridine base is too weak to completely neu-
tralize the acid and is the cause of the “sluggish” titration ob-
served in the classic KF titration reagents. If the base is too
strong, the solution becomes too alkaline and an endpoint will
not be reached. A titration in the pH range of 5–7, in which the
reaction runs quickly and stoichiometrically, is necessary. In
this method, imidazole is used as the base to neutralize these
acids and buffer the titration system. (Note: The pH of the sys-
tem could be upset if a feed sample, such as a mineral mix,
which was a strong acid or base were to be analyzed.)
The third need for the method is a totally tight titration
vessel.
Although no interferences have been encountered in the
Study Director’s laboratory with animal feed, grain, and for-
age materials, interferences are theoretically possible. One
possible source of interference is from aldehydes and ketones,
which react with methanol to form acetals and ketals, respec-
tively, and water. (For this reason, using acetone to rinse
glassware or equipment is prohibited; the acetone residue will
interfere by reacting with methanol to form acetone dimethyl
acetal and water, and thus bias water results erroneously high.)
Additional potential interferences include metal oxides, such
as CaO, or other compounds that can form water when neu-
tralized; or thiols, such as cysteine hydrochloride, or other
compounds that can be oxidized by iodine. Suspected interfer-
ences can be verified by running the water standard in tripli-
cate in the same solvent as used for the test sample immedi-
ately after the test sample, thereby retaining interfering
compounds in the titration flask. In most cases, if interferences
are present, they will affect the value for the water standard in
the same manner as the test portion was affected. Additives or
alternative solvents exist to deal with many potential interfer-
ences (e.g., 8-hydroquinone can be added to systems that are af-
fected by FeCl3). The user of the method should be alert to the
possibility of potential interferences, even if unlikely, and run
water standards frequently. Extracting replicates of a test sam-
ple in the same solvent may also magnify interference effects.
Collaborative Study
The Study Directors with the cooperation of 8 additional
participating laboratories conducted the collaborative study.
These laboratories represented a variety of laboratory types, in-
cluding research, commercial laboratories, manufacturers and
industry laboratories, and state regulatory laboratories. Two
laboratories were from outside the United States (one Canadian
and one German). Participants received no compensation. The
collaborating laboratories provided the Karl Fischer titrator, ho-
mogenizer, circulating water bath, and formamide. Karl Fischer
Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023
reagent, methanol, standard reference material, and certified
water were provided to collaborators.
Collaborating laboratories were asked to analyze 10 mate-
rials as blind duplicate pairs, resulting in 20 test samples. Test
samples were stored in airtight, heat-sealed plastic containers
to minimize moisture changes, and coded randomly. Analysts
were instructed not to open bags until immediately before
weighing test portions for analysis. Familiarization samples
were sent to each collaborator to be analyzed prior to the test
samples to acquaint them with the method and to ensure that
the laboratory was capable of handling the test samples.
Study materials were representative of animal feed, grain,
and forage materials (Table 1). Silages represent materials
high in volatile fatty acids; milk replacer represents materials
high in sugar content; urea-containing beef feed represents
urea-containing materials; and soybeans represent high fat
materials and oilseeds, etc. These are matrixes that present
problems with oven methods. All were natural or “real
world.” Approximately 20 g of each test sample material was
supplied, which was in excess of the amount needed to com-
plete the study.
Study materials were prepared as follows: Cereal grain
mix, beef feed containing urea, oats, and cat food materials
were ground to pass through a 0.75 mm screen in a Retsch Ul-
tra Centrifugal Mill ZM100 (Retsch, Hann, Germany) with a
Table 1. Collaborative test sample materials
Material Animal feed Forage Grain/oilseed
Alfalfa hay X X
Corn silage X X X
Grass hay X X
Alfalfa silage X X
Cereal grain mix X X
(birdseed, hog and
poultry ration, barley)
Oats X X
Soybeans X X (Oilseed)
Beef feed containing urea, X X X
medicated
Dry cat food X X
Milk replacer X
320 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002

Table 2. Coefficients of variation (CV) for collaborative dictions were made on triplicate test portions. Coefficients of
test samples variation (CVs) and number of values (n) are reported in Ta-
Karl Fischer NIR ble 2. In cases where one value appeared to be an outlying
Material CV (n) CV (n) value, the test was repeated. The CVs for both n = 6 (with the
questionable value removed) and for n = 7 are provided in Ta-
Alfalfa hay 1.1 (6) 0.3 (9) ble 2. An acceptance criterion of a relative standard deviation
Corn silage 2.7 (6) 1.1 (9)
(RSD) of ca 2.0% was considered acceptable for the study ma-
terials. Only the oat sample suitability was questioned. When
Grass hay 0.7 (6) 2.2 (9)
tests were repeated on a second day, the results were improved
Alfalfa silage 0.9 (6), 1.7 (7) 0.1 (9) and the test sample was retained for the collaborative study.
Cereal grain mix 0.9 (6), 2.5 (7) 0.3 (9) Collaborators were asked to report results on an “as is” ba-

Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023

(birdseed, hog and sis, to determine triplicate analysis for each test sample, and to
poultry ration, barley)
report data to 4 significant figures. In addition to data on test
Oats 4.3 (6) on 1st day 1.4 (9) on 1st day samples, collaborators were asked to return the system suit-
Oats 2.7 (6) on 2nd day 0.3 (9) on 2nd day ability and blank worksheets to the Study Directors.
Soybeans 1.5 (6), 2.3 (7) 0.4 (9)
Beef feed containing 1.6 (6) 1.1 (9) AOAC Official Method 2001.12
urea, medicated Determination of Water/Dry Matter (Moisture)
in Animal Feed, Grain, and Forage (Plant Tissue)
Dry cat food 2.4 (6) 0.3 (9)
Karl Fischer Titration Methods
Milk replacer 1.8 (6) 0.9 (9) First Action 2001
[The method is applicable to the determination of 1–15%
water in dry, ground animal feed and forage. It is not applicable
for mineral mix feed. An estimated LOD is 0.25% and an esti-
flow-through attachment. The soybeans were ground in a Foss mated LOQ is 0.8% for 500 mg test portions (estimated LOD of
Tecator Knifetec mill (Foss North America, Eden Prairie, 0.12% and estimated LOQ of 0.4% for 1g test portions).]
MN, USA). Alfalfa hay, corn silage, grass hay, and alfalfa si-
Caution: HYDRANAL®-Composite 5 contains 5 hazardous
lage materials were ground to pass through a 1.0 mm screen in components—iodine, sulfur dioxide, imidazole,
a Tecator Cyclotec mill (Foss North America). Milk replacer diethylene glycol monoethyl ether, and hydriodic
was split without grinding. No further grinding was necessary acid, and should be handled with care.
by the collaborating laboratory for any of the study materials.
Homogeneity/uniformity of the test sample sets were veri- See Tables 2001.12A–C for the results of the
fied by selecting 3 bags at random for each material and ana- interlaboratory study supporting acceptance of the methods.
lyzing them in duplicate by the proposed method in the Study
A. Principle
Director’s laboratory. The materials were also checked for ho-
mogeneity/uniformity by predicting water content for near in- For Method I, water is extracted from the animal feed or
frared (NIR) calibrations based on the proposed method. Pre- forage material into methanol–formamide (1 + 1) directly in

Table 2001.12A. Statistical results for interlaboratory study on determination of water in animal feed, cereal, and
forage (blind duplicate design)

ID Mean No. labsa(b) sr RSDr, % sR RSDR, % HORRAT

Soybeans 6.99 9 0.10 1.38 0.44 6.31 2.11
Urea feed 4.58 9 0.19 4.11 0.49 10.73 3.37
Grass hay 6.35 9 0.18 2.89 0.48 7.59 2.51
Cereal grains 8.58 9 0.39 4.49 0.46 5.35 1.85
Alfalfa silage 7.26 7(2) 0.14 1.98 0.72 9.85 3.32
Corn silage 6.40 9 0.22 3.40 0.47 7.34 2.43
Cat food 6.90 9 0.24 3.50 0.43 6.27 2.10
Milk replacer 5.21 9 0.32 6.13 0.42 7.99 2.56
Alfalfa hay 7.36 9 0.51 6.99 0.66 8.95 3.02
Oats 7.53 8(1) 0.09 1.14 0.61 8.12 2.75
a(b)
a = Number of laboratories retained after eliminating outliers, (b) = number of laboratories removed as outliers.
 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002 321

Table 2001.12B. Statistical results for collaborative study on determination of dry matter in animal feed, cereal, and
forage (blind duplicate design)
Feeds Mean Laba sr RSDr, % sR RSDR, % HORRAT

Soybeans 93.01 9 0.10 0.10 0.44 0.47 0.23

Urea feed 95.42 9 0.19 0.20 0.49 0.52 0.26
Grass hay 93.65 9 0.18 0.20 0.48 0.51 0.25
Cereal grains 91.42 9 0.39 0.42 0.46 0.50 0.25
Alfalfa silage 92.74 7(2) 0.14 0.16 0.72 0.77 0.38
Corn silage 93.60 9 0.22 0.23 0.47 0.50 0.25

Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023

Cat food 93.10 9 0.24 0.26 0.43 0.46 0.23
Milk replacer 94.79 9 0.32 0.34 0.42 0.44 0.22
Alfalfa hay 92.64 9 0.51 0.56 0.66 0.71 0.35
Oats 92.47 8(1) 0.09 0.09 0.61 0.66 0.33

a
Number of laboratories retained after eliminating outliers. Number of laboratories removed as outliers shown in parentheses.

the Karl Fischer titration vessel by high-speed homogeniza-
tion. The formamide content must not exceed 50% to maintain
reaction conditions. A subsequent titration of the water is per-
formed at 50°C with one-component Karl Fischer reagent
based on imidazole.
In Method II, boiling methanol is used instead of the meth-
anol–formamide mixture at a higher temperature of 66°C
without high-speed homogenization. A totally tight titration
vessel is critical.
These methods provide for the dry, ground test sample to
be weighed and added directly to the titration cell to minimize
contamination by atmospheric moisture. The reactions are:
CH3OH + SO2 + RN = [RNH] SO3CH3
where RN = imidazole base.
H2O + I2+ [RNH] SO3CH3 + 2RN
= [RNH] SO4CH3 + 2[RNH] I
Imidazole is used as the base to neutralize the acids and
buffer the titration system. (Note: The pH of the system could
be upset if a feed, such as a mineral mix, which was either a
strong acid or base were to be analyzed.)
One source of interference is from aldehydes and ketones,
which react with methanol to form acetals and ketals, respec-
tively, and water. Do not use acetone to rinse glassware or
equipment. The acetone residue reacts with methanol to form
acetone dimethyl acetal and water, and thus bias water results
high. Additional potential interferences include metal oxides,
such as CaO, or other compounds that can form water when
neutralized; or thiols, such as cysteine hydrochloride, or other
compounds that can be oxidized by iodine. If interferences are
suspected, verify by running the water standard in triplicate in

Table 2001.12C. Statistical results for interlaboratory study on determination of water in animal feed, cereal, and
forage, including data from laboratory using alternative extraction (blind duplicate design)
ID Mean No. labsa(b) sr RSDr, % sR RSDR, % HORRAT
Soybeans 7.01 10 0.10 1.41 0.42 6.03 2.02
Urea feed 4.57 10 0.18 3.97 0.47 10.20 3.20
Grass hay 6.34 10 0.28 4.44 0.48 7.59 2.51
Cereal grains 8.60 10 0.37 4.26 0.44 5.08 1.76
Alfalfa silage 7.40 9(1) 0.15 2.05 0.68 9.17 3.10
Corn silage 6.43 10 0.21 3.23 0.45 7.02 2.32
Cat food 6.94 10 0.23 3.33 0.44 6.27 2.10
Milk replacer 5.18 10 0.30 5.85 0.40 7.76 2.49
Alfalfa hay 7.40 10 0.49 6.64 0.64 8.59 2.90
Oats 7.51 10 0.15 1.99 0.57 7.54 2.55
a(b)
a = Number of laboratories retained after eliminating outliers, (b) = number of laboratories removed as outliers.
322 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002
the same solvent as used for the test sample immediately after
the test sample, thereby retaining interfering compounds in
the titration flask. In most cases, if interferences are present,
they will affect the value for the water standard in the same
manner as the test portion. Be alert to the possibility of potential
interferences and run water standards frequently. Extracting
replicates of a test sample in the same solvent may also magnify
interference effects. If interfering compounds are found, addi-
tives or alternative solvents exist to deal with many.
Method I
Extraction into Methanol–Formamide (1 + 1)
Using High-Speed Homogenization
B. Apparatus
(a) Karl Fischer titration-homogenization sys-
tem.—Metrohm 720 KFS Titrino, includes titration vessel LP
with water jacket (50–150 mL), “snap-in” buret unit (10 mL),
703 titration stand with pump, cable with timer
(Titrino-Polytron), or equivalent. See Figure 2001.12A
(Brinkmann Instruments, Inc., One Cantiague Rd, PO Box
1019, Westbury, NY 11590-0207, www.brinkmann.com).
(b) Circulating water bath.—Maintaining 50 ± 1°C.
(c) Homogenizer.—Brinkmann Polytron homogenizer
with PTA 20TSM foam-reducing generator with saw teeth
and knives. Assembled with generator extending into the titra-
tion vessel, and adjusted to be ca 1 in. (2.5 cm) from the bot-
tom of the titration vessel. Set to provide homogenization
speed of 24 000 rpm. See Figure 2001.12A.
(d) Glass weighing spoon.—With opening for dispensing
test portion into the titration flask through the septum stopper
(Brinkmann), or equivalent.
(e) Magnetic stirrer.
(f) Oven.—103 ± 2°C.
C. Reagents
(a) Karl Fischer reagent.—One component, based on
imidazole, with titer ca 5 mg H2O/mL reagent,
Figure 2001.12A.—Karl Fischer titration–
homogenization system.
HYDRANAL®–Composite 5 (available from Riedel-de Haën
34805, Sigma Aldrich, St. Louis, MO), or equivalent.
(b) Methanol.—Anhydrous, for moisture determinations,
water content not to exceed 0.05% (HYDRANAL–Methanol
Dry; Riedel-de Haën 34741, Sigma Aldrich, St. Louis, MO), or
equivalent.
(c) Formamide.
(d) Solvent.—Methanol–formamide (1 + 1). Mix fresh
each day.
(e) Sodium tartrate dihydrate.—Primary standard (water
content, 15.66 ± 0.05%), HYDRANAL®–standard sodium
Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023
tartrate–2-hydrate (Riedel-de Haën 34803, Aldrich Chemical
Co., Milwaukee, WI), or equivalent.
(f) Water standard.—Water standard with certificate (wa-
ter content, 10 mg/g), HYDRANAL®–water standard 10.0
(Riedel-de Haën 34849, Aldrich Chemical Co.), or equivalent.
D. Preparation of Test Sample
Grind test samples to pass a ≤1 mm opening. Mix well and
transfer to a tightly sealed container.
E. Drying or Conditioning the Cell
Dispense sufficient methanol–formamide solvent into the
titration vessel to immerse the homogenizer tip about 0.25 in.
(8 mm). Close the cell to minimize the addition of atmospheric
moisture. Heat to 50 ± 1°C. Dry the cell (including solvent,
cell walls, electrode walls, generator, and cell atmosphere) by
performing a complete run without test portion (“blank” run),
including homogenization and titration as follows: Start the
instrument and condition the solvent by titrating to remove
moisture. As soon as instrument drift has stabilized, start the
method. Homogenize at 24 000 rpm for 60 s. Enter “1” for test
portion weight, and titrate the solvent again to remove any
traces of water remaining. The end point is reached when no
change in potential is observed for 10 s (titration system pro-
grammed for stop criterion: time; delay: 10 s). A dried titration
cell has a maximum drift consumption of 5–10 µL Karl
Fischer reagent per minute.
F. Standardization
Heat cell to 50 ± 1°C. Dry the cell as in E. Depending on in-
strument, call up calibration mode. Condition solvent by titrat-
ing background moisture (hit “start”). Quickly weigh
150–250 mg of Na tartrate dihydrate standard into the glass
weighing spoon and record weight of spoon and standard to
the nearest 0.1 mg (S). Quickly transfer the weighed test por-
tion into the titration flask through the septum stopper. Ho-
mogenize at 24 000 rpm for 60 s. Reweigh empty spoon to ob-
tain tare weight (T) while homogenizing. Obtain the weight of
standard material added by subtracting tare weight (T) from
weight of spoon plus standard (S). Record weight of standard
material (S – T) in mg to the nearest 0.1 mg. Stop the homoge-
nizer after 60 s. Enter weight into instrument, turn on the stir-
rer and start the titration. Before reaching the endpoint of the
titration, homogenize momentarily to rinse down moisture
that may have been under the cap. Titrate to same endpoint as
in E, recording volume of reagent required for the titration
 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002 323

(mL reagent) in mL to the nearest 0.001 mL. Repeat 4 times. mogenize momentarily to rinse down moisture that may have
Calculate titer, then average the 5 values. The relative stan- been under the cap. This ensures that the moisture of the parti-
dard deviation should be <2%. cles under the cap or hanging on the glass walls above the liq-
uid surface is titrated. The end point is reached when no
mg H2 O mg Na 2 C4 H4 O6 ⋅ 2H2 O× 0.1566
Titer = = change in potential is observed for 10 s (stop criterion: time;
mL reagent mL reagent delay: 10 s). Record the volume of titrant (V). Repeat determi-
nation in triplicate. The relative standard deviation of repli-
where mg Na2C4H4O6⋅2H2O is S – T, in mg. cates should be <5%. The cell need not be emptied between
each titration. Usually about 3 titrations can be performed be-
G. System Suitability
fore the cell requires emptying and replenishing.
Heat cell to 50 ± 1°C. Dry the cell as in E. Check drift in the
I. Calculations

Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023

titration cell. A dried titration cell should have a maximum
drift consumption of 5–10 µL Karl Fischer reagent/min. Ana- mg H2O = V × titer
lyze a water standard as follows: Immediately after drying the
cell, break open the standard ampoule at the white ring and take V× titer
% H2O =
1–2 mL of standard with syringe which has been predried in a 10× test portion wt
103°C oven. Rinse the syringe and discard the standard solu-
tion. Draw the remaining water standard (~6 mL) into the sy- Dry matter, % = 100 – % H2O
ringe and weigh accurately by placing the syringe into a beaker
on the balance pan (SS). Quickly add ~2 mL water standard where V is the volume of titrant in mL and test portion weight
through the septum keeping the tip of the syringe below the sur- is W – T, in g.
face of the solvent. Carefully withdraw the syringe tip, reweigh
the syringe and record the weight (S1). Obtain the weight of Method II
Boiling Methanol Extraction Alternative
standard solution (Ss – S1) by subtracting the weight (S1) from
First Action 2002
the weight of the syringe plus standard (SS). Record the water
standard weight to the nearest 0.1 mg. Enter the weight into the J. Apparatus
instrument, start the stirrer, and begin the titration. Record the
(a) Karl Fischer titration system.—Metrohm 720 KFS
volume of titrant (V1). Perform the titration procedure 2 addi-
Titrino “snap-in” burette unit (10 mL), 703 titration stand, or
tional times, recording weights of the syringe after each subse-
equivalent. Incorporate a titration vessel with water jacket
quent addition (S2, S3) and the respective volume of titrant (V2,
(50–150 mL) and condenser (Horst Kühn, Am Tütberg 32,
V3). Calculate the percent recovery as follows:
3032 Fallingbostel, Germany) or a 4 neck flask with a heating
Water standard, g = SS – S1, S1 – S2, S2 – S3 mantle. See Figure 2001.12B.
(b) Circulating water bath.—Maintaining 75 ± 1°C.
Vn × titre
g standard mg / g H2 O found
Rec., % = × 100 = × 100
certified value, mg / g certified value, mg / g

Average % recovery should be 100 ± 1%. If system is not within

specifications, correct before continuing with determinations. If
the % recovery on the water standard is within specification, it is
not necessary to perform a blank run (with no material), since the
water standard indicates the condition of the system and running
a blank will provide no additional information.
H. Determination
Dry the cell as in E. Depending on the instrument call up
the sample analysis mode. Quickly weigh ~ 0.5 g test portion
(to contain ~25 to 50 mg water) into the glass weighing spoon
and record weight of the spoon plus the test portion (W).
Quickly add weighed test portion into the titration flask
through the septum stopper. Homogenize at 24 000 rpm for
60 s. Reweigh empty spoon and record tare weight (T) while
homogenizing. Obtain the test portion weight by subtracting
tare weight (T) from weight of spoon plus test portion (W).
Record weight (W – T) in g to the nearest 0.1 mg. Stop ho-
mogenizer after 60 s. Enter weight into instrument and start Figure 2001.12B.—Karl Fischer titration vessel and
the titration. Before reaching the end point of the titration, ho- condenser.
324 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002
(c) Glass weighing spoon.—With opening for dispensing
test portion into the titration flask through the septum stopper
(Brinkmann 6.2412.000 or equivalent).
(d) Magnetic stirrer.
K. Reagents
See C.
L. Preparation of Test Sample
See D.
M. Drying or Conditioning the Cell
Dispense ~50 mL methanol into the titration vessel. Close
the cell to minimize the addition of atmospheric moisture.
Heat until the MeOH begins to boil. Dry the cell (including
solvent, cell walls, electrode walls, generator, and cell atmo-
sphere) by titrating to dryness. The end point is reached when
no change in potential is observed for 10 s (titration system
programmed for “stop criterion: time; delay: 10 s). A dried ti-
tration cell has a maximum drift consumption of 5–10 µL Karl
Fischer reagent per minute.
N. Standardization
Heat cell to 66 ± 1°C (boiling point of methanol). Dry the
cell as described in M. Depending on instrument, call up cali-
bration mode. Condition solvent by titrating background
moisture (hit “start”). Switch off the heating system and when
the methanol stops boiling, quickly weigh 150–250 mg of so-
dium tartrate dihydrate standard into the glass weighing spoon
and record weight of spoon and standard to the nearest 0.1 mg
(S). Quickly transfer the weighed test portion into the titration
flask through the septum stopper. Reweigh empty spoon to
obtain tare weight (T) and obtain the weight of standard mate-
rial added by subtracting tare weight (T) from weight of spoon
plus standard (S). Record weight of standard material (S – T) in
mg to the nearest 0.1 mg. Enter weight into instrument, start the
stirrer and begin the titration. Titrate to same endpoint as de-
scribed in M, recording volume of reagent required for the titra-
tion (mL reagent) in mL to the nearest 0.001 mL. Repeat
4 times. Calculate titer as in F, then average the 5 values. The
relative standard deviation should be <2%.
O. System Suitability
Heat cell to 66 ± 1°C (boiling point of methanol). Dry the
cell as described in M. Check drift in the titration cell. A dried
titration cell should have a maximum drift consumption of
5–10 µL Karl Fischer reagent/min. Analyze a water standard
as follows: Immediately after drying the cell, switch off the
heating system and after the methanol stops an active boil,
break open the standard ampoule at the white ring and take
1–2 mL of standard with syringe which has been predried in a
103°C oven. Rinse the syringe and discard the standard solu-
tion. Draw the remaining water standard (~6 mL) into the sy-
ringe and weigh accurately by placing the syringe into a
beaker on the balance pan (SS). Quickly add ~2 mL water
standard through the septum keeping the tip of the syringe be-
low the surface of the solvent. Carefully withdraw the syringe
tip, reweigh the syringe and record the weight (S1). Obtain the
weight of standard solution (Ss – S1) by subtracting the weight
(S1) from the weight of the syringe plus test portion (SS). Re-
cord the water standard weight to the nearest 0.1 mg, and enter
the weight into instrument. Turn on the heating system, start
the stirrer and begin the titration as soon as the methanol re-
turns to an active boil. Record the volume of titrant (V1).
Perform the titration procedure 2 additional times, recording
weights of the syringe after each subsequent addition (S2, S3)
and the respective volume of titrant (V2, V3). Calculate the
percent recovery as in G.
Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023
Average % recovery should be 100 ± 1%. If system is not
within specifications, correct before continuing with determina-
tions. If the % recovery on the water standard is within specifica-
tion, it is not necessary to perform a blank run (with no material),
since the water standard indicates the condition of the system and
running a blank will provide no additional information.
P. Determination
Dry the cell as described in M. Depending on the instru-
ment call up the sample analysis mode. After switching off the
heating system and the methanol stops an active boil, quickly
weigh ~0.5 g test portion (to contain ~25 to 50 mg water) into
the glass weighing spoon and record weight of the spoon plus
the test portion (W). Quickly add weighed test portion into the
titration flask through the septum stopper. Reweigh empty
spoon and record tare weight (T). Obtain the test portion
weight by subtracting tare weight (T) from weight of spoon
plus test portion (W). Record weight (W – T) in g to the near-
est 0.1 mg. Enter weight into instrument, start the stirrer, turn
on the heating system, and begin the titration as soon as the
methanol returns to an active boil. The end point is reached
when no change in potential is observed for 10 s (stop criterion:
time; delay: 10 s). Record the volume of titrant (V). Repeat de-
termination in triplicate. The relative standard deviation of rep-
licates should be <5%. The cell need not be emptied between
each titration. Usually about 3 titrations can be performed be-
fore the cell requires emptying and replenishing.
Q. Calculations
See I.
Ref.: J. AOAC Int. 85, 320–324(2002)
Results and Discussion
Study materials were shipped in April 1999. Results were
received from 9 laboratories (Table 3) over 19 months, with
the last set being received in November 2000. Contact was lost
with a 10th laboratory, and no data were received from it.
Equipment for this method is specialized and was not avail-
able in most laboratories before onset of the study. Few col-
laborators had prior experience with the equipment or the
method. Therefore, a number of unanticipated problems oc-
curred in collaborating laboratories, and set-up times were
longer than anticipated in some laboratories. Collaborating
laboratories reported results on familiarization samples before
testing the test samples. Results of their performance on famil-
 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002 325

7.03a

7.46b
iarization samples were reported to them, and attempts were

6.86

7.37
8.11
6.67

7.87
8.52
7.55
7.38
B19
made to correct problems at that point. However, in some

Oats
cases many months elapsed between the time some laborato-

7.23a

8.00b
6.84

7.20
8.24
6.73

7.82
8.39
7.41
7.55
A17
ries analyzed the familiarization samples and the test samples.
Unfortunately, since the concentration of water can be af-
fected by atmospheric conditions in the laboratory after the

7.63a
7.29

7.21
8.19
6.38

7.57
7.88
6.83
7.09
6.37
B20
Alfalfa hay

sample bag is opened, only initial data could be submitted, and

no laboratories had an opportunity to correct problems and
7.87a reanalyze the test samples. Despite the number of problems
7.70

6.87
7.89
6.28

7.40
8.09
7.02
8.38
7.98
A16

and time over which data were submitted, results of the collab-
orative study are encouraging. Eight laboratories used

Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023

Metrohm/Brinkmann (Westbury, NY) equipment and one
4.93a
4.76

4.87
5.69
4.91
Milk replacer

5.29
5.61
4.84
5.08
4.69
B18

laboratory (Laboratory 5) used equipment manufactured by

Thermo Orion (Beverly, MA). The Orion equipment required
4.96a
5.00

4.85
5.68
4.93

5.52
5.67
4.81
5.86
5.74
A14

a different standardization procedure than specified in this

method; therefore, Laboratory 5 used the water standard as the
primary standard, and sodium tartrate dihydrate to check sys-
7.45a
7.58

6.53
6.72
7.06

7.17
6.69
6.87
7.36
5.96
B15

tem suitability.
Cat food

A laboratory ranking using the test described in Youden

and Steiner (4) detected some bias among participating labo-
7.29a
7.44

6.56
6.77
7.04

7.06
7.46
6.63
6.90
6.34
A11

ratories. Overall, Laboratory 5 ranked lowest (reported the

highest water values) with a laboratory ranking score of 22
(Table 4). It could be interpreted that the Orion equip-
6.72a
6.65

6.28
6.81
5.81

6.66
6.66
6.22
6.81
5.44
B13
Corn silage
Table 3. Collaborative study results for determination of water in animal feed, cereal, and forage

ment/procedure is not equivalent to the Brinkmann equip-

ment/procedure; however, Laboratory 8, which used the
6.63a
6.47

6.15
6.97
5.66

6.21
6.94
6.50
6.98
6.05
A9

Brinkmann equipment also reported relatively high water val-

Water, %

ues. Laboratory 8 was one of the last laboratories to report

study results, so the storage or handling of the study materials
5.42b,c 4.21b,c
8.13a 7.79a

7.81b 7.72b
8.35

7.14
7.47
7.39

6.78
7.55
5.87
Alfalfa silage
B12

may also be questioned. Since it was not possible to sort bias

due to the method or technique from bias due to handling of
the test samples, and since a minimum number of laboratories
8.50

7.08
7.52
7.35

6.91
7.43
6.34
A8

Data excluded by Cochran’s test when excluding 100% boiling methanol extraction data.

participated in the study, all laboratories were included in

Data excluded by Cochran’s test when including 100% boiling methanol extraction data.

study statistics.
Tables 2001.12A and B show a summary of results of sta-
8.80a
Cereal grains

8.74

8.22
8.84
8.62
8.95
8.01
8.59
9.04
7.67
B10

tistical analysis of study data. The within-laboratory RSDr (re-

peatability) ranged from 1.14 to 6.99% for water or from 0.09
8.70a
8.80

8.05
8.19
8.59
9.56
9.08
8.36
8.79
8.43

to 0.56% for dry matter. Among-laboratory relative (including

A6

within-) RSDR (reproducibility) ranged from 5.35 to 10.73%,

or from 0.44 to 0.77% for dry matter. The Study Directors feel
6.76a

that this method performance is acceptable, considering the

5.92

6.01
6.97
6.31
6.13
6.05
6.01
7.23
6.72
B7
Grass hay

inexperience of the collaborators and the length of time for

study results to be reported to the Study Directors. The RSDr,
5.77a
5.62

6.03
6.92
6.28
6.68
5.92
6.03
7.07
6.32
A3

RSDR, and HORRAT values are consistent with other KF

methods determining water at concentrations <15% mois-
ture (5–8). Because water concentration of test samples
4.53a
4.37

4.29
4.92
4.57
5.63
4.33
4.12
5.23
4.26
B4

changes as material is opened and exposed to atmosphere and

Urea feed

laboratory environment, reruns to correct errors were impossi-

100% Boiling methanol extraction.
4.39a

ble. Laboratory performance would be expected to improve

4.35

4.33
5.00
4.54
5.07
3.78
4.14
5.16
4.37
A2

with additional experience with the method and equipment.

Comparability Data for Alternative Extraction
7.31a
7.41

6.70
6.99
6.91
7.76
6.54
6.62
7.18
6.72
B5
Soybeans

Laboratory 1 provided comparability data for an alterna-

tive to the proposed solvent in an attempt to eliminate
7.14a
7.28

6.63
7.14
6.97
7.86
6.54
6.49
7.47
6.63
A1

formamide. The alternate solvent falls into a category of “pro-

cedure modification.” A paired t-test showed no significant
difference between the 2 extraction procedures (p = 0.1982).
Lab

A laboratory ranking using the test described in Youden and

1
1
2
3
4
5
6
7
8
9

c
326 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002
Table 4. Comparison of laboratory ranking scores including and excluding alternative extraction procedure data
Titration
Lab equipment used Excluding data using alternative extraction
1 Metrohm/Brinkmann 701 Titrino
1 Metrohm/Brinkmann 701 Titrino
2 Metrohm/Brinkmann 716 DMS Titrino
3 Metrohm/Brinkmann 784 KF Titrino
4 Metrohm/Brinkmann 716 DMS Titrino
5 Orion Turbo 2
6 Metrohm/Brinkmann 701 KF Titrino
7 Metrohm/Brinkmann 720 KFS Titrino
8 Metrohm/Brinkmann 758 KDF Titrino
9 Metrohm/Brinkmann 758 KDF Titrino
a
Boiling methanol extraction.
b,c
Showed statistical bias compared to other laboratories based on laboratory ranking score.
Steiner (4) showed that the alternative-extraction procedure
had no bias compared to the original extraction procedure.
The boiling methanol extraction procedure had a laboratory
ranking score of 44. Results of that test are shown in Table 4.
Comparability data for the study including the boiling
methanol extraction procedure are provided in Table 3. The
within-laboratory relative standard deviation (RSDr, repeat-
ability) ranged from 1.41 to 6.64% when including the 100%
boiling methanol data compared to 1.14 to 6.99% when ex-
cluding it. Among-laboratory (including within-) relative
standard deviation (RSDR, reproducibility) ranged from 5.08
to 10.20% when including the 100% boiling methanol data
compared to 5.35 to 10.73% when excluding it. Including
100% boiling methanol data did not affect collaborative study
results (see Table 2001.12C). The values obtained with the
boiling methanol extraction were not detected as outliers for
laboratory variance by Cochran’s test (4).
Recommendations
On the basis of this study, the Study Directors recommend
that the described method for Determination of Water (Mois-
ture) in Animal Feed, Grain, and Forage (Plant Tissue) by
Karl Fischer Titration method be adopted as Official First Ac-
tion. Due to the high bias associated with the method modifi-
cation for the Orion Turbo 2 equipment, this modification
should be further evaluated before being adopted as Official
First Action. It is recommended that the alternative boiling
methanol extraction procedure be adopted Official First Ac-
tion in addition to the original method extraction procedure.
Acknowledgments
We thank Riedel-de Haen for providing Karl Fischer re-
agent, methanol, standard reference material, and certified
Laboratory ranking score
Including data using alternative extraction
— 44a
44 50
71.5 80.5
31 35
61 69
Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023
22b 25c
74b 82c
56.5 64.5
b
24 26c
66 74
water to collaborators and for coming to our rescue in the ship-
ment of these materials. We thank Helga Hoffmann for
persisting with the alternative extraction and for generating
the comparability data. We thank Brian Steinlicht (South Da-
kota State University, Olson Biochemistry Laboratories,
Brookings, SD) for assistance with preparing, splitting, and
sealing familiarization and test samples. We also thank the
following collaborators for their participation in this study:
Jacques A. Boucher and Andy Clark, Canadian Food In-
spection Agency, Laboratory Services Division, Ottawa, On-
tario, Canada
Doug Clark, Riedel-de Haen, St. Louis, MO
David Dow and Pam Fahey, 3M, Maplewood, MN
Helga Hoffmann, Rdh Laborchemikalien Gmbh & Co.
KG, HYDRANAL-Labor, Seelze, Germany
Dave Mertens and Diane Amundson, U.S. Dairy Forage
Research Center, Madison, WI
Herschel Morris and Paul Coco, Louisiana Department of
Agriculture and Forestry, Division of Agricultural Chemistry,
Baton Rouge, LA
Tina Rittenhouse, Hershey Foods, Hershey, PA
Myriam Wohl, Brinkmann Instruments, Inc., Westbury, NY
References
(1) Van Erem, T., Thiex, N., Pohmer, J., Poffenbarger, M.,
Smith, V., & Patel, E. (1998) J. AOAC Int. 81, 25–32
(2) Riedel-de Haën HYDRANAL®-Manual, available through
Allied Signal, HYDRANAL® Technical Center, Morristown, NJ
(3) Scholz, E. (1984) Karl Fischer Titration: Determination of
Water-Chemical Laboratory Practice, Springer-Verlag, New
York, NY
(4) Youden, W.J. & Steiner, E.H. (1975) Statistical Manual of
the AOAC, Association of Official Analytical Chemists,
Arlington, VA
 THIEX & VAN EREM: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 2, 2002 327

(5) Horwitz, W., Albert, R., Deutsch, M.J., & Thompson, N.

(1990) J. AOAC Int. 73, 661–680
(6) Bernetti, R., Kochan, S.J., & Pienkowski, J.J. (1984) J.
Assoc. Off. Anal. Chem. 67, 299–301
(7) Martin, R.A., Jr (1977) J. Assoc. Off. Anal. Chem. 60,
654–657
(8) Duncan, R.D. (1972) J. Assoc. Off. Anal. Chem. 55, 699–701

Downloaded from https://academic.oup.com/jaoac/article/85/2/318/5656498 by Varendra University user on 15 June 2023