Bartos et al.: Journal of AOAC International Vol. 97, No.

3, 2014  687

SPECIAL GUEST EDITOR SECTION

Determination of Phosphorus and Potassium in Commercial

Inorganic Fertilizers by Inductively Coupled Plasma-Optical
Emission Spectrometry: Single-Laboratory Validation
James M. Bartos
Office of Indiana State Chemist, 175 South University St, West Lafayette, IN 47907
Barton L. Boggs, J. Harold Falls, and Sanford A. Siegel
CF Industries, 10608 Paul Buchman Hwy, Plant City, FL 33565

A two-part single-laboratory validation study was methods use one or more chemical reagents that have safety
conducted for determination of the P and K content concerns, including quinoline (benzopyridine), formaldehyde,
in commercial fertilizer materials by inductively tetraphenylboron, and perchloric acid. An added benefit is that
coupled plasma-optical emission spectrometry (ICP- inductively coupled plasma-optical emission spectrometry
OES). While several methods exist for determination (ICP-OES) instruments can provide simultaneous determination
of P and K in fertilizer products, the main focus of of both P and K, whereas most existing methodologies require
this study was on ICP-OES determination, which the separate determination of each. The instrument analysis time
offers several unique advantages. Fertilizer samples for this method is approximately 2  min/sample. The primary
with consensus P and K values from the Magruder waste generated by this method is excess sample extract
and Association of Fertilizer and Phosphate solution, which can be disposed of safely in the laboratory
Chemists (AFPC) check sample programs were environment, and only requires basic personal protective
selected for this study. Validation materials ranging equipment. Many laboratories engaged in fertilizer testing have
from 4.4 to 52.4% P2O5 (1.7 to 22.7% P) and 3 to 62% ICP-OES instrumentation; however, there is not an AOAC
K2O (2.5 to 51.5% K) were utilized. Because all P approved method for P and K determination by ICP-OES.
and K compounds contained in fertilizer materials
are not “available” for plants to use, this study Principle
was conducted in two parts. Part A focused on
ammonium citrate–disodium EDTA as the extraction P and K are macronutrients, or nutrients required by plants in
solvent, as it estimates the pool of fertilizer P and higher concentrations (2). Not all sources of P and K contained
K that is considered available to plants. Part B in fertilizer products are considered readily available for plant
focused on hydrochloric acid as the digestion use. Differentiating between the plant “available” and “total”
solvent, as it estimates the total P and K content of nutrients requires the use of different extraction solvents.
the fertilizer product. Selectivity studies indicated After solubilizing the desired form of fertilizer P and K in
that this method can have a high bias for fertilizer their respective solvents, the resulting solution is analyzed.
products containing sources of phosphite or organic An ICP-OES instrument utilizes a plasma or high-temperature
P compared to gravimetric or colorimetric methods energy source to ionize or excite electrons to higher energy
that measure just orthophosphate. Provided the orbitals, which return to ground state releasing characteristic
analytical challenges outlined in this study are UV or visible wavelengths that can be quantified (3). This study
addressed, this method offers the potential for a is presented in two parts, A and B, which represent the two types
quick, accurate, and safe alternative for determining of fertilizer P and K commonly tested.
the P and K content of commercial inorganic
fertilizer materials. Part A: Ammonium Citrate–Disodium EDTA-Soluble P
and K

S
Study Design
everal AOAC methods (935.02, 949.01, 955.06, 957.02,
958.01, 958.02, 960.02, 960.03, 962.02, 969.02, 969.04, This portion of the single-laboratory validation (SLV) study
971.01, 978.01, and 983.02) exist for the determination is applicable to the determination of the ammonium citrate–
of P and K in fertilizer materials (1). Historically, these methods disodium EDTA-soluble P, termed “direct available phosphate,”
have worked well, but most utilize slower, more labor-intensive and K, termed “soluble potash,” by ICP-OES. A number of
methodologies, including gravimetric, titrimetric, photometric, method challenges were encountered. The main challenges
and colorimetric techniques. Furthermore, several of these that the direct available phosphate and soluble potash present
are: the high carbon content in the citrate–EDTA solvent that
Guest edited as a special report from the AOAC Agricultural can deposit in the torch, the carbon chains can reduce plasma
Community on “Collaborations in New and Improved Methods of
Analysis for Plant Food Materials” by Nancy Thiex.
efficiency, thereby requiring a higher plasma energy for P, and
Corresponding author’s e-mail: [email protected] the dynamic K concentration range requires the use of multiple
DOI: 10.5740/jaoacint.12-399 wavelengths and possible sample dilution. These challenges

Copyright: © 2014 AOAC INTERNATIONAL.

This is an open-access article, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
688  Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014

Table  1.  Validation materials and their P and K content, expressed as P and P2O5 and as K and K2O
Available P, Available P2O5, Total P, Total P2O5, Soluble K, Soluble K2O,
a
Grade Source % % % % % %

34-0-0 ACS reagent grade 0 0 0 0 0 0

21-0-0 ACS reagent grade 0 0 0 0 0 0

5-5-5 Magruder 2006-05 1.93 4.42 2.45 5.60 2.62 3.16

10-10-10 Magruder 2005-10 4.34 9.93 4.38 10.04 8.63 10.39

19-19-19 Magruder 2006-03 8.38 19.20 8.27 18.94 14.98 18.05

6-24-24 Magruder 2004-06 9.77 22.38 9.89 22.64 21.19 25.53

18-46-0 Magruder 2005-01 20.34 46.60 20.47 46.88 NA NA

11-52-0 AFPC 2006-10 22.85 52.33 22.86 52.36 NA NA

0-0-50 Magruder 2001-10 NA NA NA NA 42.33 50.99

0-0-60 Magruder 2005-12 NA NA NA NA 51.58 62.13
a
  No total K values are documented for these check sample materials.

are reduced by performing an “online” dilution, achieved Table  1. Some benefits of using Magruder and AFPC check
by using a small sample pump tube and a large ionic buffer/ samples as validation materials are that they are preground and
internal standard pump tube, which dilutes the matrix and homogeneous, have consensus values for multiple methods,
sample concentration approximately four-fold. If the method and are immediately ready for use after minimal mixing. Check
challenges listed are properly addressed, the direct available samples representing several common fertilizer materials
phosphate and soluble potash content in commercial inorganic with low, medium, and high levels of P and K were chosen.
fertilizers, in the absence of organically bound P and K and of
3– Validation materials were stored in sealed glass jars for short-
phosphite (PO3 ), in the concentration ranges of 0.06 to 53%
term use and in heat sealed plastic pouches at room temperature
P2O5 (0.026 to 23% P) and 0.01 to 63% K2O (0.008 to 53% K)
can be determined. for long-term storage. Negative control material, namely ACS
reagent grade ammonium nitrate (34-0-0) and ACS reagent
Validation Materials grade ammonium sulfate (21-0-0) were utilized.

Validation materials were selected from fertilizer materials A.  Apparatus (Part A)
maintained by the Magruder (www.magruderchecksample.org)
(4) and Association of Fertilizer and Phosphate Chemist (AFPC; (a)  Analytical balance.—Readability to 0.1 mg, AT 200
www.afpc.net) (5) check sample programs, and are listed in (Mettler Toledo, Columbus, OH).

Table  2.  ICP calibration standards from stock reagent salts for citrate–EDTA-soluble P and K
P2O5 P2O5 K2 O K2O
Volume, Citrate, Stock 1, Stock 2, P concn, P2O5, solution, sample, K concn, K2O, solution, sample,
a b
Standard ID mL mL mL mL µg/mL µg/mL % % µg/mL µg/mL % %

Blank 1000 400 0 0 0 0 0 0 0 0 0 0

1 250 100 10 of Std 7 NA 12 27.5 0.00275 1.4 15.15 18.25 0.00182 0.9

2 250 100 20 of Std 7 NA 24 55 0.00550 2.7 30.3 36.5 0.00365 1.8

3 250 100 5 NA 50 115 0.01146 5.7 63.1 76 0.00760 3.8

4 250 100 10 NA 100 229 0.02291 11.5 126 152 0.01521 7.6

5 250 100 15 NA 150 344 0.03437 17.2 189 228 0.02281 11.4

6 250 100 22 NA 220 504 0.05041 25.2 278 335 0.03345 16.7

7 250 100 30 NA 300 687 0.06874 34.4 379 456 0.04562 22.8

8 250 100 40 NA 400 917 0.09165 45.8 505 608 0.06083 30.4

9 250 100 50 NA 500 1146 0.11457 57.3 631 760 0.07603 38

10 250 100 NA 25 NA NA NA NA 747 900 0.08998 45

11 250 100 NA 30 NA NA NA NA 897 1081 0.10805 54

12 250 100 NA 35 NA NA NA NA 1046 1260 0.12600 63
a
  S
 tock 1 = 2500 μg/mL P stock standard = 2.7461 g potassium dihydrogen phosphate (KH2PO4)/250 mL prepared in deionized water. Also, 10 of Std 7
represents a serial dilution from another standard, meaning take 10 mL from standard 7.
b
  Stock 2 = 7472 μg/mL K stock standard = 3.5615 g potassium chloride or 4.8299 g potassium nitrate/250 mL in deionized water.
 Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014  689

Table  3.  Calibration criteria for “direct-available” P and “soluble” K by ICP-OES

Element ID Wavelength, nm a Calibration range, μg/mL Standards used (see Table 2) Curve fit Spectral deconvolution

P 177.434 (1) 0 to 100 Blank, 1, 2, 3, 4 Linear None

P 177.434 (2) 100 to 500 4, 5, 6, 7, 8, 9 Linear None
P 178.222 (1) 0 to 100 Blank, 1, 2, 3, 4 Linear None
P 178.222 (2) 100 to 500 4, 5, 6, 7, 8, 9 Linear None
P 213.618 (1) 0 to 100 Blank, 1, 2, 3, 4 Linear Cu 213.598
P 213.618 (2) 100 to 500 4, 5, 6, 7, 8, 9 Linear Cu 213.598
P 214.914 (1) 0 to 100 Blank, 1, 2, 3, 4 Linear Cu 214.898
P 214.914 (2) 100 to 500 4, 5, 6, 7, 8, 9 Linear Cu 214.898
K 766.485 (1) 0 to 126 Blank, 1, 2, 3, 4 Quadratic None
K 766.485 (2) 50 to 379 3, 4, 5, 6, 7 Quadratic None
K 769.897 126 to 505 4, 5, 6, 7, 8 Quadratic Interference LiNO3
K 404.721 505 to 1046 8, 9, 10, 11, 12 Quadratic None
a
  (1) and (2) distinguish between the same wavelength used to cover two separate concentration ranges.

(b)  pH meter.—Readability to pH 7.00, Model 8005 (VWR (i)  Triton X-100.—Octyl phenol ethoxylate (J.T. Baker
Scientific, Radnor, PA). Chemicals).
(c)  pH combination electrode.—Orion 9102BNWP (Thermo (j)  10 000 μg/mL Be standard.—In 4% HNO3, product
Fisher Scientific, Waltham, MA). 10M5-1 (High Purity Standards, Charleston, SC).
(d)  Constant temperature water bath.—Capable of (k)  10 000 μg/mL Sc standard.—In 4% HNO3, product
maintaining bath temperatures of 65 ± 2°C, BK53, Yamato
10M48-1 (High Purity Standards).
Scientific, Santa Clara, CA.
(l)  Cesium chloride.—CsCl, FW 168.36, purity > 99.999%
(e)  Heated shaking water bath.—Capable of maintaining
(Sigma-Aldrich, St. Louis, MO).
bath temperatures of 65 ± 2°C, and set to approximately
200 reciprocations/min, designed internally by the Purdue (m)  Lithium nitrate.—LiNO3, FW 68.95, purity > 99%, (EM
University Central Machine Shop (West Lafayette, IN). Science, Gibbstown, NJ).
(f)  ICP-OES instrument.—Vista Pro-Axial view (Agilent (n)  Ammonium nitrate.—NH4NO3, FW 80.04, ACS grade,
Technologies, Santa Clara, CA). purity > 99% (Mallinckrodt Chemicals).
(g)  Gated riffle splitter.—Model 16-25X (Carpco Inc., (o)  Ammonium sulfate.—(NH4)2SO4, FW 132.14, ACS
Jacksonville, FL). grade, purity > 99% (Sigma-Aldrich).
(h)  Sample grinder.—Model ZM200 rotor mill (Retsch, (p)  Citrate–EDTA extraction solution.—A pH neutral 0.11 M
Haan, Germany). ammonium citrate and 0.033  M disodium EDTA solution is
prepared by adding 25  g disodium EDTA and 50  g dibasic
B. Reagents (Part A)
ammonium citrate to approximately 1500  mL deionized (or
equivalent) water contained in a 2 L volumetric flask. The pH is
(a)  Ammonium citrate, dibasic.—(NH4)2HC6H5O7,
adjusted to near neutral by adding 30 mL ammonium hydroxide–
formula weight (FW) 226.19, ACS grade, purity >98% (EMD
Chemicals, Darmstadt, Germany). water (1 + 1, v/v) solution in a fume hood. The final adjustment
(b)  EDTA, disodium salt, dihydrate.—C10H14N2Na2O8·2H2O, to pH 7.00 (±0.05) is achieved using a pH electrode and meter,
FW 372.24, purity > 99% (J.T. Baker Chemicals, Center Valley, and adding the ammonium hydroxide–water (1 + 1, v/v) solution
PA). drop-by-drop. After a stable pH of 7.00 (±0.05) is achieved, the
(c)  Ammonium hydroxide.—NH4OH, FW 35.05, 28.0–30.0% flask is brought to volume with deionized water and mixed.
as NH3 (Mallinckrodt Chemicals, Center Valley, PA). This solution can be prepared in larger volumes; however, it is
(d)  Nitric acid.—HNO3, 67–70%, OmniTrace grade (EMD susceptible to microbial activity resulting in a maximum shelf
Chemicals). life of 2 weeks if stored in a dark location.
(e)  Potassium dihydrogen phosphate.—KH2PO4, (q)  0.5% Triton-X.—Add 1 mL Triton X-100 [(i) above] to
certified at 22.73% P and 28.73% K, National
a 200 mL volumetric flask and dilute to volume with deionized
Institute of Standards and Technology (NIST)  200a
water.
(Gaithersburg, MD), www.nist.gov/srm.
(r)  Internal standard/ionization buffer for citrate–EDTA-
(f)  Potassium chloride.—KCl, FW 74.55, ACS grade, purity
>99% (Mallinckrodt Chemicals). soluble P and K.—A solution of 10 µg Sc/mL in 0.018 M CsCl
(g)  Potassium nitrate.—KNO3, certified at 38.66% K, NIST and 4% nitric acid was prepared by adding 1  mL 10 000  μg
193. Sc/mL stock standard [(k) above], 3  g cesium chloride [(l)
(h)  Sodium phosphate, dibasic anhydrous.—Na2HPO4, FW above], 20 mL nitric acid [(d) above], and 1 mL 0.5% Triton-X
141.96, ACS grade (Mallinckrodt Chemicals). [(q) above]/liter. If Be is used as an internal standard, add 4 mL
690  Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014
Figure 1. Example of multiple wavelengths and splitting a wavelength into calibration segments to cover
the 0 to 1000 ug K/ml range.

however, this will require a secondary dilution of samples with

a high concentration of K. Diluted samples must maintain the
solvent matrix, which can be prepared by diluting 400  mL
citrate–EDTA extract solution [Part A, B(n)] to 1 L.
The sample extract and internal standard/ionic buffer
solutions were blended using a Y-connector (Part  30703-90;
Cole-Parmer, Bunker, CT) just prior to the nebulizer using the
conditions described in Table 4.

D. Sample Preparation (Part A)

The validation materials obtained from the Magruder (3) and

AFPC (4) check sample programs were prepared according to
their internal grinding and homogeneity protocols. A portion
of these materials was poured into a 9  oz. glass jar that was
thoroughly mixed by rotation prior to weighing. Additional
validation materials, used for method comparison, were
prepared internally. These materials were prepared by riffling
approximately 2300 g (or 5 lb) of field-collected samples down
to approximately 150 g with a gated riffle splitter [Part A, A(g)],
Figure  1.  Example of multiple wavelengths and splitting a and the subsample was ground with a rotor mill [Part A, A(h)]
wavelength into calibration segments to cover the 0 to 1000 µg K/mL
range. to pass a 0.75 mm mesh screen.

of 10 000  µg/mL Be [(j) above] stock standard to obtain a E. Extraction (Part A)

concentration of 40 µg/mL Be.
A 0.5 ± 0.01  g fertilizer test portion was weighed and
C. Calibration (Part A) completely transferred to a 250  mL wide-mouth class A
volumetric flask. Next, 100  mL 65 ± 2°C preheated citrate–
(a)  Standard solution.—Stock standards were made from EDTA extraction solution [Part A, B (p)] was dispensed into
potassium dihydrogen phosphate [Part A, B(e)], potassium each flask, followed by insertion of a rubber stopper. The test
chloride [Part A, B(f)], and potassium nitrate [Part A, B(g)]. solutions were shaken in a 65 ± 2°C preheated water bath set
Several calibration standards are required for the following to approximately 200  rpm for exactly 60  min; then removed
reasons: multiple ICP-OES wavelengths were utilized; some from the water bath, allowed to cool to room temperature (20 to
wavelengths were split into multiple calibration segments; and 25°C), brought to volume with deionized (or equivalent) water,
a minimum of five points/curve is recommended. The standards stoppered, and mixed. Any test solution (or extract) containing
used and their P and K concentrations, expressed as µg/mL and suspended debris was filtered using P and K free filters. These
as their percentage oxide forms, are listed in Table 2. An attempt
to use commercially available stock standard solutions was Table  4.  Final ICP-OES conditions utilized for citrate–
made; however, these resulted in low P recoveries, possibly due EDTA-soluble P and K validation
to the matrix change resulting from the acid used to preserve
Factor Setting
the standards. Therefore, using standards prepared only from
potassium dihydrogen phosphate [Part A, B(e)], potassium Power, kW 1.45
chloride [Part A, B(f)], or potassium nitrate [Part A, B(g)], as Plasma flow, L/min 19.5
outlined in Table 2, is recommended. Auxiliary flow, L/min 2.25
(b) ICP-OES calibration.—Emission intensity for each of
Nebulizer pressure, L/min 0.7
the calibration standards was plotted against concentration.
A minimum of five calibration standards is recommended Nebulizer type Seaspray
for each wavelength. An internal standard [Part A, B(r)] is Spray chamber Cyclonic
used to correct the concentration of the calibration standards,
Sample pump tube Black/black
as well as the samples. The wavelengths, standards used,
Buffer/internal standard pump tube Gray/gray
concentration ranges, curve fit, and neighboring wavelengths
that may have spectral overlap are listed in Table 3. Whenever CsCl ionic buffer concn, M 0.018
possible, linear regression is preferred. Because of the dynamic Internal standard and concn, µg/mL 10
range in fertilizer K concentration, quadratic curve fit may be
Buffer matrix 4% nitric acid
necessary, provided the curvature is not excessive. Many newer
ICP software programs have algorithms to detect excessive Exposure length, s 10

curvature of second order or quadratic calibration curves. An Number of exposures 3

example of the curves used for the citrate–EDTA-soluble K Rinse time, s 35
is shown in Figure  1. Alternatively, linear calibration can be
Total sample analysis time, min/sample 2
achieved by removal of the high concentration K standards;
 Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014  691
Figure 2. Difference in ammonium citrate-EDTA soluble % P2O5 and K2O concentration
for the validation materials as a result of changes in optimization variables

Table  5.  Ruggedness factors for citrate–EDTA-soluble P

and K
No. Optimization factor Major value Minor value

1 Power, kW A = 1.40 a = 1.50

2 Nebulizer flowa, L/min B = 0.70 b = 0.80

3 Exposure length, s C = 10 c = 20

4 Internal standard D = Sc d = Be

5 Rinse time, s E = 30 e = 40

6 CsCl buffer concn, M F = 0.018 F = 0.024

7 Buffer matrix G = 4% nitric acid g = water

a
  Cyclonic spray chamber with concentric glass nebulizer. Figure  2.  Difference in ammonium citrate–EDTA-soluble P2O5 and
K2O concentration (%) for the validation materials as a result of
changes in optimization variables.

test solutions (or extracts) have a very limited shelf life, so

analysis occurred within 16 h of extraction. Alternatively, the standards can be entered with their
theoretical percentage P2O5 and K2O in solution values, listed
in Table 1.
F. ICP Conditions (Part A)

Optimization testing was conducted using the Youden Method Validation: Citrate-EDTA-Soluble P and K
ruggedness trial (6) to determine the effects of seven variables:
power, nebulizer flow, exposure length, internal standard type, A. Accuracy/Bias (Part A)
rinse time, CsCl buffer concentration, and buffer matrix. The Three separate extractions were analyzed on 4 different
major and minor criteria are listed in Table 5. Because AOAC days, resulting in a total of 12  individual accuracy readings
Method 993.31 addressed the citrate–EDTA solvent extraction for each test sample. Results for accuracy are provided in
process, extraction conditions were not included as part of the Table 6. Recoveries for P ranged from 99.3 to 124.9%. A closer
ruggedness testing.
examination of the high recovery for the Magruder  2006-05
As a general indication of the method ruggedness, the mean
test sample suggests that it contains appreciable organic P. As
data difference in the major versus minor variables for all
a result, there is a potential for a high bias with this method
validation materials is listed in Figure 2. Relative to the P and
for fertilizers containing organic P. Removal of the organic
K concentration of the test samples, there was little difference
P sample improved the recovery range to 99.3 to 101.7%.
in the data between the major and minor variables, suggesting
Recoveries for K ranged from 98.4 to 100.7%. With the
the method is sufficiently rugged over the ranges tested.
However, there is some sensitivity to rinse time and buffer exception of the Magruder 2006-05 test sample that contained
concentration, so these conditions should be closely monitored. organic P, recoveries of the consensus or certified P values were
The final instrument conditions used for method validation of well within the recommended range (6).
citrate–EDTA-soluble P and K are listed in Table 4. Another indicator of method performance and potential
bias was assessed by comparing observed data with data from
G. Calculations (Parts A and B) other recognized methods. For P, 160  commercial inorganic
fertilizer samples, which were tested by the proposed ICP-OES
Several options exist in the software for data reporting, method, were also tested by AOAC  993.31 (automated
including expression of standard concentrations, units, test spectrophotometer). For K, 83 commercial inorganic fertilizer
portion weight, test solution volume, and dilution factor. The samples, which were tested by the proposed ICP-OES
calibration standards are prepared as µg/mL of P and K, and method, were also tested by AOAC  983.02 (automated flame
the final fertilizer results reported as percentage P2O5 and K2O, photometry). The data comparisons are offered in Figure  3.
which requires the following calculation: Statistical analysis by the t-test at P = 0.05 indicated the results
by all methods were not statistically different.
P2O5 ,% = [P × (250/W) × 142/(31 × 2)]/10 000
B. Precision (Part A)
where P = ICP-OES P reading in µg/mL, 250 = final volume
in mL, W = test portion weight in g, 142 = MW of P2O5, Three separate extractions were analyzed on 4 different
31 = MW of P, 2 = mole ratio of P2O5/P, and 10 000 = conversion
days, resulting in a total of 12  individual precision readings
of µg/mL to %.
for each test sample. The mean, SD, RSD, Horwitz predicted
K2O, % = [K × (250/W) × 94.2/(39.1 × 2)]/10 000 RSD (PRSD), and HorRat (r) repeatability values are listed in
Tables 7 and 8. In some instances, the RSD was considerably
where K = ICP-OES K reading in µg/mL, 250 = final volume in below the PRSD, resulting in HorRat (r) values below the
mL, W = test portion weight in g, 94.2 = MW of K2O, 39.1 = MW suggested range of 0.3 to 1.3 (7). This is possibly due to the fact
of K, 2 = mole ratio of K2O/K, and 10 000 = conversion of that the validation materials used were check sample materials,
μg/mL to %. which are uncharacteristically uniform, and that the high P and
692  Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014

Table  6.  Recoveries of citrate–EDTA-soluble P and K for all validation materials

Sample Grade Obs P, %a Cons P, %a P Rec, % Obs K, %a Cons K, %a K Rec, %
b
Magruder 2006-05 5-5-5 5.52 4.42 124.9 3.15 3.16 99.7
Magruder 2005-10 10-10-10 9.93 9.93 100.0 10.30 10.39 99.1

Magruder 2006-03 19-19-19 19.21 19.20 100.1 18.18 18.05 100.7

Magruder 2004-06 6-24-24 22.76 22.38 101.7 25.13 25.53 98.4

c
Magruder 2005-01 18-46-0 46.62 46.60 100.0 NA NA NA

AFPC 2006-10 11-52-0 53.03 52.36 101.3 NA NA NA

Magruder 2001-10 0-0-50 NA NA NA 50.90 50.99 99.8

Magruder 2005-12 0-0-60 NA NA NA 62.26 62.13 100.2

NISTd 200a KH2PO4a NA 51.76 52.11 99.3 34.73 34.64 100.3

a
  P and K values listed are based on P2O5 and K2O content. Obs = observed value, Cons = consensus value, and Rec = recovery.
b
  Magruder 2006-05 contains 58% organic matter by loss-on-ignition.
c
  NA = Not applicable.
d
  NIST = National Institute of Standards and Technology.

K concentration in fertilizers produce robust signals that are very probability for spectral overlap occurs when the fertilizer Cu
easily quantified and replicated by ICP-OES instrumentation. concentration is high and the P concentration is low, which is
unusual but possible. This situation can largely be corrected
C.  Selectivity (Part A) by using spectral deconvolution or interelement correction
modeling, which is found in most of the instrument software.
Fortified reagent blank and standard addition studies were Wavelengths 177.434 and 178.222  nm for P are also an
conducted to evaluate the matrix effect or interactions. For the option on some ICP-OES instruments, and are relatively free
fortified blank study, four treatments were prepared: an unspiked from spectral interference. However, these lines are weaker
reagent blank, a low range concentration of 20  µg P/mL and than the 213.618 and 214.914  nm wavelengths, which could
50  µg K/mL, a medium range concentration of 200  µg P/mL impact recovery of some of the lower P concentrations. For K,
and 500  µg K/mL, and a high range concentration of 400  µg wavelength 769.897 nm showed signs of spectral interference
P/mL and 1000 µg K/mL. The four treatments were performed from lithium nitrate. The interference was too strong to correct
in triplicate and were measured on 4 separate days. Recoveries with spectral deconvolution or interelement correction. There
for the fortified blanks ranged from 99.6 to 101.2% and are are two options available to avoid this problem: use CsCl
listed in Table  9. For the standard addition study, a midlevel rather than LiNO3 as the ionic buffer, or avoid the use of the
test sample, namely Magruder 2004-06 with a value of 22.38% wavelength 769.897  nm. Whenever possible, using multiple
P2O5 (or 195 µg/mL P) and 25.53% K2O (or 424 µg/mL K) was wavelengths is encouraged for result confirmation, or so that an
spiked with 10, 30, and 130% of its P and K concentration from unusual result can be identified and corrected.
sodium phosphate dibasic and potassium chloride, respectively.
These treatments were performed in triplicate and measured on D.  LOD and LOQ (Part A)
4 separate days. Recoveries of P and K for the standard addition
study ranged from 98.91 to 102.44% and are listed in Table 10. Reagent blank test solutions were prepared separately and
Some selectivity issues were identified during method measured in triplicate on 4  separate days. The mean blank
development. The phosphate forms considered “available” reading plus three times the SD was used to calculate the LOD,
to plants are primary and secondary orthophosphate, or while the mean blank reading plus 10 times the SD was used to
H2PO4– and HPO42–, respectively  (8). The gravimetric and Figure 3. Comparison of results from 160 fertilizer samples for P2O5 content by ICP-OES and AOAC
spectrophotometric methods react with and specifically method 993.31 and from 83 fertilizer samples for K2O by ICP-OES and AOAC method 983.02.

measure only orthophosphate. ICP-OES, however, measures

all ionizable forms of P found in the fertilizer extract, so it
will measure orthophosphate plus any other form of P that is
soluble in the fertilizer extract solution. As a result, there are
some known instances in which this method can produce a high
P bias. Two examples are given in Figures  4 and  5, namely
phosphite, or PO33–, and organic P. As a result, this method
is not suitable for organic fertilizers and fertilizers containing
phosphite. A few instances of spectral overlap or interference
are also known and are listed in Table 3, with an example
provided in Figure  6. The P wavelengths with the greatest
Figure  3.  Comparison of results from 160 fertilizer samples for
sensitivity are 213.618 and 214.914  nm; however, both of P2O5 content by ICP-OES and AOAC Method 993.31, and from
these lines have a potential for Cu interference. The greatest 83 fertilizer samples for K2O by ICP-OES and AOAC Method 983.02.
 Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014  693

Table  7.  Repeatability precision data for citrate–EDTA- Table  8.  Repeatability precision data for citrate–EDTA-
soluble P by ICP-OES soluble K by ICP-OES
Mean, RSD, PRSD, HorRat Mean, RSD, PRSD, HorRat
Sample Grade % SD % % (r) Sample Grade % SD % % (r)

Magruder 2006-05 5-5-5 5.52 0.04 0.70 3.09 0.23 Magruder 2006-05 5-5-5 3.15 0.02 0.64 3.37 0.19
Magruder 2005-10 10-10-10 9.93 0.13 1.30 2.83 0.46 Magruder 2005-10 10-10-10 10.30 0.16 1.52 2.82 0.54

Magruder 2006-03 19-19-19 19.21 0.14 0.72 2.56 0.28 Magruder 2006-03 19-19-19 18.18 0.21 1.17 2.59 0.45

Magruder 2004-06 6-24-24 22.76 0.15 0.67 2.50 0.27 Magruder 2004-06 6-24-24 25.13 0.26 1.02 2.46 0.42

Magruder 2005-01 18-46-0 46.62 0.20 0.44 2.24 0.19 Magruder 2001-10 0-0-50 50.90 0.21 0.41 2.21 0.19
AFPC 2006-10 11-52-0 53.03 0.15 0.28 2.20 0.13 Magruder 2005-12 0-0-60 62.26 0.54 0.87 2.15 0.40

calculate the LOD (6). With these criteria, the LOD based upon medium, and high levels of P and K were selected to ensure
the P2O5 and K2O content in fertilizer were 0.0014 and 0.003% the method performed well over a dynamic concentration range.
respectively. The LOQ based upon the P2O5 and K2O content
in fertilizer were 0.06 and 0.01%, respectively. As a result, the A.  Apparatus (Part B)
method is applicable to 0.06 to 53% P2O5 (or 0.026 to 23% P)
and 0.01 to 63% K2O (or 0.008 to 53% K). (a)  Balance.—Readability to 0.1 mg, Sartorius BP210S
(Gottingen, Germany).
Part B: Acid-Soluble P and K (b)  Hot plate.—Model 53015, Lindburg/Blue M (Watertown,
WI).
Study Design (c)  ICP-OES instrument.—Thermo 6500 Duo View (Thermo
Scientific, Cambridge, UK).
This portion of the SLV study is applicable to the determination (d)  Gated riffle splitter.—SP-177 Jones Standard Aluminum
of the hydrochloric “acid-soluble” or the presumed “total” P Splitter (Gilson Co., Inc., Lewis Center, OH).
and K content in fertilizer materials by ICP-OES. While the (e)  Sample grinder.—Mikro-Samplemill (Pulverizing
total P and K concentrations found in a fertilizer material may Machinery, Summit, NJ).
not all be available for plant use, there are several reasons
why this determination is necessary. Some examples include: B.  Reagents (Part B)
determination of the P and K content of mined or processed
ores; a more productive alternative to AOAC methods 957.02, (a)  Hydrochloric acid.—HCl, 35–38%, trace metal grade,
958.01, 962.02, 969.02, or 978.01  (9); and determination of Cat. No. A508-500 (Fisher Scientific, Pittsburgh, PA).
“available” P by the indirect method, namely AOAC 960.02 (9). (b)  Ammonium dihydrogen phosphate.—NH4H2PO4, FW
Hydrochloric acid may not completely digest or release all 115.03, trace metal basis, purity >99.999%, Cat. No. 204005-
organically bound P, so this method should be limited to 100G (Sigma-Aldrich).
inorganic fertilizer materials. For inorganic fertilizer materials, (c)  Potassium chloride.—KCl, FW 74.55, trace metal basis,
acid-soluble concentrations of 0.05 to 53% P2O5 (0.022 to 23% purity > 99.99%, Cat. No. 204099-250G (Sigma-Aldrich).
P) and 0.3 to 63% K2O (0.25 to 53% K) can be determined by (d)  Scandium oxide.—Sc2O3, FW 137.91, Item No. OX21-
this method. 5N (Stanford Materials Corp., Irvine, CA).
(e)  Nitric acid.—HNO3, 69.2%, certified ACS plus grade,
Validation Samples Cat. No. A200 C212 (Fisher Scientific).
(f)  Triton X-100.—Polyethylene-glycol p-tert-octylphenyl
The validation samples, listed in Table 1, were also used ether, 4-(C8H17)C6H4(OCH2CH2)nOH (n approximately 10),
for Part B of this study. Common fertilizer materials with low, FW 624, Cat. No. BP151-500 (Fisher Scientific).

Table  9.  Ammonium citrate–disodium EDTA matrix blank solution spiked with low, medium, and high concentrations of P
and K, and their recoveries
Blank + spike concn, Concn in fertilizer, Replicate 1, Replicate 2, Replicate 3, Mean recovery,
Sample ID ppm % ppm ppm ppm %

Fortified low P 20 2.3 P2O5 20.06 20.11 20.02 100.30

Fortified medium P 200 23 P2O5 202.6 202.1 202.9 101.26

Fortified high P 400 46 P2O5 401.3 400.9 404.0 100.52

Fortified low K 50 3 K2O 49.66 50.84 49.77 100.19

Fortified medium K 500 30 K2O 501.3 506.8 507.3 101.03

Fortified high K 1000 60 K2O 993.7 999.9 996.5 99.67

694  Bartos et al.: Journal of AOAC International Vol. 97,Figure
No. 3,5. 2014
Recovery of “available phosphate” from a fertilizer source
containing organic P by the proposed and other reference methods.
Table  10.  Standard addition recoveries from a mid-level P
6
and K fertilizer sample spiked with 10, 30, and 130% of its P
and K concentration
Standard addition P Recovery, % K Recovery, % 5

6-24-24 + 2.4% P2O5 + 2.4% K2O 102.44 98.96

4
6-24-24 + 7.2% P2O5 + 7.2% K2O 101.97 98.91

6-24-24 + 31.2% P2O5 + 31.2% K2O 101.86 99.90

% P2O5
3

(g)  Cesium chloride.—CsCl, FW 168.36, trace metal basis, 2

purity >99.999%, Cat. No. 203025-50G (Sigma-Aldrich).

(h)  Lithium nitrate.—LiNO3 ReagentPlus grade, FW 68.95, 1

Cat. No. 227986-1KG (Sigma-Aldrich).

(i)  10 000 μg/mL Be stock standard.—In 5% HNO3, Cat. No. 0

PLBE-10-500 (Exaxol Corp., Clearwater, FL). Magruder 2006-05

(j)  10 000 μg/mL Sc stock standard.—Weigh 15.3374  g Group Grand Average Citrate-EDTA ICP

scandium oxide [(d) above] into a 600 mL beaker. Add 300 mL Figure  5.  Recovery of “available phosphate” from a fertilizer
deionized water, then slowly add 100 mL nitric acid [(e) above]. source containing organic P by the proposed and other reference
methods.
Heat solution on a hotplate to a gentle boil, and continue boiling
until the solution becomes clear.
hydrochloric acid [Part B, B(a) above]. Bring to volume with
(k)  1% Triton X.—Pipet 10 mL Triton X-100 solution [(f)
deionized water and mix well.
above] into a 1 L flask. Bring to volume with deionized water
and mix well. C.  Calibration (Part B)
(l)  Internal standard/ionization buffer for total P and K.—A
60  μg Sc/mL in 0.035 M CsCl and 2% HNO3 solution is (a)  Standard solution.—Stock standards were made from
prepared by adding 6 mL 10 000 µg Sc/mL stock standard [(j) ammonium dihydrogen phosphate [Part B, B(b)] and potassium
above], 6  g CsCl [(g) above], 20  mL HNO3 [(e) above], and chloride [Part B, B(c)]. Again, many calibration standards are
2 mL 1% Triton X [(k) above] per L. If LiNO3 is used as the required for several reasons: multiple ICP-OES wavelengths were
ionic buffer, replace the CsCl with 8 g LiNO3 [(h) above]. Also, utilized; some wavelengths were split into multiple calibration
segments; and a minimum of five points/curve is recommended.
if Be is used as an internal standard, add 1 mL 10 000 µg/mL
The standards used and their P and K concentrations expressed as
Be stock standard solution [(i) above] to obtain a 10 µg/mL Be
µg/mL and as their percentage oxide forms are listed in Table 11.
internal standard concentration. An attempt to use commercially available stock standard
(m)  Hydrochloric acid digestion solution.—4 M. Into a 1 L solutions was made; however, better P recoveries were obtained
volumetric flask filled approximately half full with deionized or using weighed salts.
equivalent water, slowly add a total of 333 mL of concentrated Figure 6. Spectral overlap of Cu wavelength
Figure 4. Recovery of “available phosphate” from a phosphite fertilizer 214.898 nm into P wavelength 214.914 nm.
source by the proposed and other reference methods.

20

18

16

14
% P2O5

12

10

Magruder 2007-12B
AOAC 960.03 Gravimetric AOAC 993.31 Spectrophotometric Citrate-EDTA ICP

Figure  4.  Recovery of “available phosphate” from a phosphite Figure  6.  Spectral overlap of Cu wavelength 214.898 nm into P
fertilizer source by the proposed and other reference methods. wavelength 214.914 nm.
 Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014  695

Table  11.  ICP-OES calibration standards from stock reagent salts for “total” P and K
P2O5, P2O5, K2O K2O
Standard Vol., Acid, Weight Weight P concn, P2O5, solution, sample, K concn, K2O, solution, sample,
a
ID mL mL NH4H2PO4, g KCl, g µg/mL µg/mL % % µg/mL µg/mL % %

Blank 1000 40 0 0 0 0 0 0 0 0 0 0
b
1 1000 40 40 of Standard 6 0.6305 9.8 22.4 0.00224 1 332 400 0.0400 20

2 1000 36 100 of Standard 10b 0.4748 47 108 0.01076 5 249 300 0.0300 15

3 500 12 100 of Standard 10b 100 of Standard 14b 94 215 0.02153 11 163 196 0.0196 10

4 1000 32 0.4539 200 of Standard 12b 122 280 0.02802 14 116 140 0.0140 7

5 1000 36 0.6810 100 of Standard 14b 184 420 0.04204 21 81 98 0.0098 5

b
6 1000 40 0.9079 50 of Standard 13 245 561 0.05605 28 34.9 42 0.0042 2

7 1000 40 1.1349 25 of Standard 13b 306 701 0.07007 35 17.4 21 0.0021 1

c
8 1000 40 1.3619 NA 367 841 0.08408 42 NA NA NA NA

9 1000 40 1.5888 NA 428 981 0.09809 49 NA NA NA NA

10 1000 40 1.7510 NA 472 1081 0.10811 54 NA NA NA NA

11 1000 40 NA 0.7915 NA NA NA NA 415 500 0.0500 25

12 1000 40 NA 1.1079 NA NA NA NA 581 700 0.0700 35

13 1000 40 NA 1.3295 NA NA NA NA 697 840 0.0840 42

14 1000 40 NA 1.5511 NA NA NA NA 814 980 0.0980 49

15 1000 40 NA 1.7727 NA NA NA NA 930 1120 0.1120 56

16 1000 40 NA 1.9943 NA NA NA NA 1046 1260 0.1260 63

17 1000 40 0.9728 0.9497 262 601 0.06006 30 498 600 0.0600 30

a
  Acid = Volume of 1:2 HCl–water required to make standard.
b
  Serial dilution from another standard (e.g., 40 of Standard 6 means add 40 mL of Standard 6).
c
  NA = Not applicable.

(b)  ICP-OES calibration.—Emission intensity for each of one or more of the top K standards listed in Table  11 and/or
the calibration standards was plotted against concentration. conducting sample dilutions using 0.16 M HCl.
A minimum of five calibration standards is recommended for The sample extract and internal standard/ionic buffer
each wavelength. An internal standard [Part B, B(l)] is used to solutions were blended using a reagent “T” connector (Part
correct the concentration of the calibration standards as well as No. 116-0522-01, Bran+Luebbe, Mequon, WI) just prior to the
the samples. The wavelengths, standards used, concentration nebulizer using the conditions described in Table 13.
ranges, curve fit, and neighboring wavelengths that may require
spectral deconvolution are listed in Table  12. The data in D.  Sample Preparation (Part B)
Table 12 are based on a radial view for K. If linear regression to
1000 µg/mL of K is not possible, one or more of the following The validation materials obtained from the Magruder (4) and
may be necessary: quadratic curve fit (provided the curvature AFPC (5) check samples programs were prepared according to
is not excessive), utilizing wavelength 404.721 nm for the five their internal grinding and homogeneity protocols. Additional
highest K calibration standards listed in Table  11, dropping validation materials, used for method development and data

Table  12.  Calibration criteria for acid-soluble or “total” P and K

Element ID Wavelength, nma Calibration range, µg/mL Standards used (see Table 2) Curve fit Spectral deconvolution

P 213.618 (1) 0 to 245 Blank, 1, 2, 3, 4, 5, 6 Linear Cu 213.598

P 213.618 (2) 184 to 472 5, 6, 7, 8, 9, 10 Linear Cu 213.598

P 214.914 (1) 0 to 245 Blank, 1, 2, 3, 4, 5, 6 Linear Cu 214.898

P 214.914 (2) 184 to 472 5, 6, 7, 8, 9, 10 Linear Cu 214.898

K 766.485 (1) 0 to 332 Blank, 1, 2, 3, 4, 5, 6, 7 Linear None

K 766.485 (2) 332 to 1046 11, 12, 13, 14, 15, 16, 17 Linear None

K 769.897 (1) 0 to 332 Blank, 1, 2, 3, 4, 5, 6, 7 Linear Possible LiNO3

K 769.897 (2) 332 to 1046 11, 12, 13, 14, 15, 16, 17 Linear Possible LiNO3
a
  (1) and (2) distinguish between the same wavelength used to cover two separate concentration ranges.
696  Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014

Table  13.  Final ICP-OES conditions utilized for acid-soluble Table  14.  Optimization factors for acid-soluble or “total”
or “total” P and K validation P and K
Factor Setting No. Optimization factor Major value Minor value

Power, kW 1.15 1 Test portion weight, g A = 0.55 a = 0.45

Plasma flow, L/min 15 2 Volume 4 M HCl, mL B = 11 b=9

Auxiliary flow, L/min 1.5 3 Digestion time, min C = 17 c = 13

Nebulizer pressure, L/min 0.40 4 ICP power, kW D = 1.2 d = 1.1

Nebulizer type V-grove 5 Internal standard E = Sc e = Be

Spray chamber Scott’s (baffled) 6 Ionic buffer F = Li F = Cs

Sample pump tube Orange/white (id = 0.64 mm) 7 Nebulizer flow, L/min a G = 0.42 g = 0.38

Buffer/internal standard pump tube Orange/white (id = 0.64 mm) a

  V-grove nebulizer with baffled spray chamber.
CsCl concentration, M 0.035

Internal standard and concn, µg/mL 6

major and minor variables are listed in Table 14. As a general
Buffer matrix 2% nitric acid indication of method ruggedness, the mean data difference in
Exposure length, s 10 the major versus minor variables for all the test materials is
Number of exposures 3 listed in Figure 7. Lower test portion weights produced slightly
lower results, so limited deviation (±0.025 g) from a test portion
Rinse time, s 30
weight of 0.5 g is recommended. Also, K is sensitive to nebulizer
Total sample analysis time, min 2.4
pressure/flow, which should be closely monitored, as nebulizers
can deteriorate over time. Relative to the sample concentration,
there was little difference in the data between the major and
comparisons, were prepared internally. These materials were minor variables for all other parameters, suggesting the method
prepared by riffling [Part B, A(d)] the entire 2300 g (or 5 lb) field- is sufficiently rugged. The final instrument conditions used for
collected sample three times before further splitting the sample method validation of acid-soluble or “total” P and K are listed
down to approximately 100 g. The entire 100 g subsample was in Table 13.
then ground [Part B, A(e)] to pass a Tyler-35 mesh sieve (U.S.
Standard Sieve Size No. 40, 0.420 mm or 0.165 in. opening; G.  Calculations (Part B)
Fisherbrand, stainless steel; Fisher Scientific, Pittsburgh, PA).
The calculations listed in Section G (Part A) also apply to
E.  Extraction (Part B) acid-soluble or “total” P and K.

For total P and K, a 0.5 ± 0.01 g test portion was weighed Method Validation: Acid-Soluble P and K
and completely transferred to a 250  mL class A volumetric
flask. Next, 30 mL deionized (or equivalent) water was added,
A.  Accuracy/Bias (Part B)
followed by addition of 10  mL HCl–water (1 + 2, v/v) acid
digestion solution [Part B, B(m)]. The samples were placed on Three separate extractions were analyzed on four different
a preheated hotplate that resulted in a gentle boil for 15 ± 1 min. days, resulting in a total of 12  individual accuracy readings
The volumetric flasks were allowed to cool to room temperature for each test sample. Results for accuracy are provided in
(20 to 25°C); then brought to volume with deionized water. The Table  15. Recoveries for P ranged from 95.53 to 99.40%. A
acid strength of the final digestion solution was approximately closer examination into the recovery for the Magruder 2006-05
0.04  M HCl. Any sample containing suspended debris was Figure 7. Difference in acid-soluble % P2O5 and K2O concentration for the test samples as
filtered using P and K free filters. These samples have a limited a result of changes in optimization variables.

shelf life, so all sample analysis should occur within 2 weeks

of digestion. After repeated heating and cooling cycles of the
volumetric flasks, they may lose calibration. Routinely, the
calibration of the 250 mL volumetric flasks should be checked
by adding 250 g deionized (or equivalent) water and verifying
the volume is at the meniscus. If a flask loses calibration, either
use the corrected volume established by water weight, or safely
discard the flask.

F.  ICP-OES Conditions (Part B)

Optimization testing was conducted using the Youden

ruggedness trial (6) to determine the effects of seven variables:
Figure  7.  Difference in acid-soluble P2O5 and K2O concentration
test portion weight, solvent volume, digestion time, ICP power, (%) for the test samples as a result of changes in optimization
internal standard type, ionic buffer type, and nebulizer flow. The variables.
 Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014  697

Table  15.  Acid-soluble or “total” P and K recoveries for the validation samples
Source Grade Obs P, %a Cons P, %a P Rec, % Obs K, %a Cons K, %a K Rec, %
b
Magruder 2006-05 5-5-5 5.35 5.60 95.53 3.39 3.16 107.28 b
Magruder 2005-10 10-10-10 9.71 10.04 97.10 10.22 10.39 98.36

Magruder 2006-03 19-19-19 18.71 19.03 98.32 17.79 18.05 98.56

Magruder 2004-06 6-24-24 22.44 22.64 99.12 25.88 25.53 101.37

Magruder 2005-01 18-46-0 46.60 46.88 99.40 NAc NA NA

Magruder 2001-10 0-0-50 NA NA NA 50.46 50.99 98.96

Magruder 2005-12 0-0-60 NA NA NA 62.26 62.13 100.21

a
  P and K values listed are based upon P2O5 and K2O content. Obs = observed, Cons = consensus value, and Rec = recovery.
b
  Magruder 2006-05 contains 58% organic matter by loss-on-ignition.
c
  NA = Not applicable.

test sample suggests that it contains appreciable organic P. ranging from 97.10 to 100.45% for total P and 97.16 to 102.38%
Hydrochloric acid may not completely digest or release all P for K.
from organic compounds, and therefore, could result in a low
bias. Also, the types of P extracted from organic compounds may B.  Precision (Part B)
3–
be different from phosphate (PO4 ), which could create a bias
relative to methods (i.e., gravimetric and spectrophotometric) Three separate extractions were analyzed on 4  different
that react solely with orthophosphate. As a result, there is a days, resulting in a total of 12  individual precision readings
potential for a bias with this method for fertilizers containing for each test sample. The mean, SD, RSD, PRSD, and HorRat
organic P. Removal of the organic sample improved the (r) repeatability values are listed in Tables 17 and 18. In most
recovery range to 97.10 to 99.40%. Similarly, with removal instances, the HorRat (r) values were in, or near, the suggested
range of 0.3 to 1.3 (7).
of the organic sample, recoveries for K ranged from 98.36 to
101.37%. With the exception of the Magruder  2006-05 test
C.  Selectivity (Part B)
sample that contained organic materials, recoveries of the
consensus values were within the recommended range (6).
Fortified reagent blank and standard addition studies were
Another indicator of method performance and potential bias
conducted to evaluate matrix effect or interactions. For the
was assessed by comparing observed data with consensus data
fortified blank study, four treatments were prepared: an unspiked
from other recognized methods. Because the check sample
reagent blank, a low range concentration of 20  µg P/mL and
validation materials had been tested by multiple methods, 50 µg K/mL, a medium range concentration of 200 µg P/mL and
comparisons of the data recovery with other common methods 500 µg K/mL, and a high range concentration of 400 µg P/mL
are listed in Table 16. Statistical analysis (t-test, P = 0.05) and 1000  µg K/mL. The four treatments were performed in
indicated there was no statistical difference in the results triplicate, and were measured on 4 separate days. Recoveries
obtained for the methods listed in Table  6. However, the for the fortified blanks ranged from 98.82 to 100.12% and are
recoveries listed in Table 16 reinforce that the candidate method listed in Table 19. For the standard addition study, a midlevel
may produce a bias for organic fertilizer materials. For common test sample, namely Magruder 2004-06, with a value of 22.64%
inorganic materials, this proposed method produced recoveries P2O5 (or 198 μg/mL P) and 25.53% K2O (or 424 µg/mL K), was

Table  16.  Comparison of acid-soluble P and K results to other recognized “total” P and soluble K methods
Source Grade P Rec, 962.02, %a P Rec, 958.01, %a K Rec, 958.02, %a K Rec, 958.02, %a

Magruder 2006-05 5-5-5 117.58b 118.36b 111.51b 107.96b

Magruder 2005-10 10-10-10 97.20 97.10 97.89 97.71

Magruder 2006-03 19-19-19 98.42 98.22 98.08 97.16

Magruder 2004-06 6-24-24 100.18 100.45 100.98 100.31

Magruder 2005-01 18-46-0 99.12 100.45 NAc NA

Magruder 2001-10 0-0-50 NA NA 99.27 102.38

Magruder 2005-12 0-0-60 NA NA 99.52 99.94

a
 P and K values listed are based upon P2O5 and K2O content. 962.02 = AOAC gravimetric method; 958.01 = AOAC spectrophotometric method;
958.02 = AOAC volumetric tetraphenylboron method; and 938.02 = flame photometric method. Rec = recovery.
b
  Magruder 2006-05 contains 58% organic matter by loss-on-ignition.
c
  NA = Not applicable.
698  Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014

Table  17.  Repeatability precision data for acid-soluble or “total” P by ICP-OES

Sample Grade Mean, % SD, % RSD, % PRSD, % HorRat (r)

Magruder 2006-05 5-5-5 5.35 0.06 1.12 3.09 0.36

Magruder 2005-10 10-10-10 9.71 0.11 1.13 2.83 0.40

Magruder 2006-03 19-19-19 18.71 0.21 1.12 2.56 0.44

Magruder 2004-06 6-24-24 22.44 0.25 1.11 2.50 0.45

Magruder 2005-01 18-46-0 46.60 0.33 0.71 2.24 0.32

spiked with 10, 30, and 130% of its P and K concentration from materials over a wide range of P and K concentrations to
ammonium dihydrogen phosphate [Part B, B(b)] and potassium ensure acceptable method performance. For ammonium-
chloride [Part B, B(c)], respectively. These treatments were citrate–disodium EDTA-soluble P and K, it is critical that
performed in triplicate and measured on 4  separate days. three aspects of this method be maintained, namely, a small
Recoveries of P and K for the standard addition study ranged sample pump tube and a large ionic buffer/internal standard
from 99.23 to 101.65%, and are listed in Table 20. pump tube as noted in Table 4 to dilute the carbon content of
The selectivity concerns mentioned for ammonium-citrate the sample matrix, a dilute ionic strength buffer solution to
disodium EDTA-soluble P and K in Section C (Part A) also apply prevent “salting” of the torch, and a high instrument power
to acid-soluble P and K by ICP-OES. Specifically, as shown in setting to overcome the reduction in plasma efficiency resulting
Tables 15 and 16 for the Magruder 2006-05 validation sample, from the carbon chains contained in the citrate–EDTA solvent.
this proposed method can have a bias for fertilizer materials The higher power setting requires a corresponding increase in
containing organic sources of P and K. As a result, this method plasma coolant and auxiliary flow settings to prevent the glass
should be limited to fertilizer materials containing inorganic torch from melting. Failure to maintain all the aforementioned
sources of P and K. Likewise, as demonstrated in Figure 6 and criteria greatly reduces the quality of the data. Any instrument
Table 12, selection of wavelengths free of spectral interference modification must not result in data variation that exceeds the
or deconvolution modeling may be required in instances where investigational allowance listed in the Association of American
the fertilizer material contains Cu and/or LiNO3. Plant Food Control Officials official publication  (9) for a
reference sample of known concentration. Low P recoveries
D. LOD and LOQ (Part B) will occur if the matrix is not sufficiently diluted using the
proper pump tube ratio, the plasma power setting is too low,
Blank solutions of 0.16 M HCl were prepared separately or the extraction solvent and calibration standard matrix is not
and measured in triplicate on 4 separate days. The mean blank properly matched. The greatest challenge with K is splitting
reading plus three times the SD was used to calculate the LOD, the dynamic K solution concentration range of 0 to 1000  µg
while the mean blank reading plus 10 times the SD was used to K/mL into appropriate segments, as demonstrated in Figure 1
calculate the LOQ (6). With these criteria, the LOD based upon and Tables  3 and  12. As a result, to measure the range of K
the P2O5 and K2O content in fertilizer were 0.012 and 0.120%, expected in fertilizer materials, one or more of the following
respectively. The LOQ based upon the P2O5 and K2O content in will be required: radial view, quadratic curve fit (provided the
fertilizer were 0.05 and 0.30%, respectively. curvature is not excessive), utilizing wavelength 404.721  nm
for the five highest K calibration standards listed in Tables  2
Results and Discussion and 11, dropping one or more of the top K standards listed in
Tables 2 and 11, and/or conducting sample dilutions. Because
This represents an advanced ICP-OES method, so the 404.721 nm intensity is much weaker than the 766.485 and
familiarization, practice and demonstration of proficiency before 769.987  nm intensities, if used, the 404.721  nm wavelength
use is strongly recommended. Also, differences in ICP-OES should be closely monitored for signal strength and K recovery.
instrument designs, introduction systems, plasma efficiency, If any minor, performance-based changes are made to the
viewing orientation, radio frequency (RF) generation, etc., could pump tube ratios listed in Tables 4 and 13, the corresponding
necessitate some minor modifications. Any minor modification Be and Sc internal standard concentrations and ionic CsCl or
must be performance based, using a variety of known reference LiNO3 buffer concentrations should be modified to maintain

Table  18.  Repeatability precision data for acid-soluble or “total” K by ICP-OES

Sample Grade Mean, % SD, % RSD, % PRSD, % HorRat (r)

Magruder 2006-05 5-5-5 3.39 0.04 1.18 3.36 0.35

Magruder 2005-10 10-10-10 10.22 0.08 0.80 2.81 0.29

Magruder 2006-03 19-19-19 17.79 0.07 0.39 2.59 0.15

Magruder 2004-06 6-24-24 25.88 0.17 0.66 2.45 0.27

Magruder 2001-10 0-0-50 50.46 0.33 0.65 2.22 0.29

Magruder 2005-12 0-0-60 62.26 0.56 0.90 2.15 0.42
 Bartos et al.: Journal of AOAC International Vol. 97, No. 3, 2014  699

Table  19.  3 M HCl blank solution spiked with low, medium, and high concentrations of P and K, and their recoveries
Sample ID Blank + spike concn, ppm Concn in fertilizer, % Replicate 1 Replicate 2 Replicate 3 Mean recovery, %

Fortified low P 20 2.3 P2O5 19.72 19.97 19.89 99.30

Fortified medium P 200 23 P2O5 197.0 199.5 196.4 98.82

Fortified high P 400 46 P2O5 396.4 401.2 397.9 99.63

Fortified low K 50 3 K 2O 49.90 50.28 50.00 100.12

Fortified medium K 500 30 K2O 498.7 500.7 498.9 99.89

Fortified high K 1000 60 K2O 997.5 1003.3 998.5 99.97

configuration and instrument conditions was also received from

Table  20.  Standard addition recoveries from a mid-level P
Doug Shrader with Agilent Technologies.
and K fertilizer sample spike with 10, 30, and 130% of its P
and K concentration
References
Standard addition P recovery, % K recovery, %

6-24-24 + 2.4% P2O5 + 2.4% K2O 100.80 101.65   (1) Official Methods of Analysis (2005) 18th Ed., AOAC
INTERNATIONAL, Gaithersburg, MD
6-24-24 + 7.2% P2O5 + 7.2% K2O 99.23 101.08
  (2) Brady, N.C., & Weil R.R. (2008) The Nature and Property of
6-24-24 + 31.2% P2O5 + 31.2% K2O 100.29 101.08 Soils, 14th Ed., Pearson-Prentice Hall, Upper Saddle River, NJ
420–438
  (3) Xiandeng, H., & Jones, B.T. (2000) Encyclopedia of Analytical
the concentrations listed in reagent sections Part A [B(r)] and
Chemistry, R.A. Meyers (Ed.), John Wiley & Sons Ltd,
Part B [B(l)]. Chichester, UK, pp 9468–9485
Provided the aforementioned challenges are addressed,   (4) Magruder Fertilizer Check Sample Program. www.
this method offers several unique advantages, including an magruderchecksample.org (accessed September 6, 2012)
  (5) Association of Fertilizer and Phosphate Chemists. www.afpc.net
elimination of unsafe chemicals and waste products, lower (accessed September 6, 2012)
detection capabilities, a dynamic working range, a reduction in   (6) Horwitz, W. (2002) Requirements for Single-Laboratory
labor, and an increase in productivity. Validation of Chemical Methods, Draft 2002-11-07. http://www.
aoac.org/Ag_Materials/additives/aoac_slv.pdf
  (7) Horwitz, W. (2004) AOAC 2004 HORRAT_SLV.pdf. http://
Acknowledgments www.aoac.org/dietsupp6/Dietary-Supplement-web-site/
HORRAT_SLV.pdf (accessed July 16, 2012)
We thank Connie Lehe, Sally Mullins, Natalie Newlon, and   (8) Schachtman, D.P., Reid, R.J., & Ayling, S.M. (1998) Plant
Physiol. 116, 447–453. http://dx.doi.org/10.1104/pp 116.2.447
Irene Szalasny with the Office of the Indiana State Chemist
  (9) Association of American Plant Food Control Officials
for their dedicated assistance and support of the citrate–EDTA (AAPFCO), Official Publication No. 65 (2012) p. 191. www.
portion of this validation study. Valuable advice on ICP-OES aapfco.org