3.3 Nitrogen Determination by Combustion Method
Protein (Crude) in Animal Feed: Combustion Method. (990.03)
Official Methods of Analysis. 1990. Association of Official
Analytical Chemists. 15th Edition.
This method is applicable for the determination of nitrogen in
all types of forages.
Nitrogen freed by combustion at high temperature in pure oxygen
is measured by thermal conductivity detection and converted to
equivalent protein by appropriate numerical factor.
Any instrument or device designed to measure nitrogen by
combustion may be used which is equipped to provide following
- Furnace to maintain minimum operating temperature of
950oC for pyrolysis of sample in pure (99.9%) oxygen.
Some systems may require higher temperature.
- Isolation system to isolate liberated nitrogen gas from
other combustion products for subsequent measurement by
thermal conductivity detector. Device for converting NO2
products to N2 or measuring N as NO2 may be required and
included in the system.
- Detection system to interpret detector response as %
nitrogen (weight/weight). May include features such as
calibration on standard material, blank determination and
barometric pressure compensation. Any required
calibration must be based on theoretical % nitrogen in
pure primary standard organic material such as NIST SRM
Uric Acid 913 or EDTA.
- Follow manufacturer's recommendation for safe operation
- Secure compressed gas cylinders and use proper gas
Operate instrument according to manufacturer's instructions;
following are generalized instructions.
- Turn furnaces on (or take off standby).
- Turn gas regulators to desired flow rate.
- Wait until furnaces have stabilized at desired
- Enter sample number on console.
- Enter other parameters as required by computer software.
- Enter appropriate N content of pure primary standard.
- Include two blanks and three dried or dessicated pure
primary standards at the beginning of each run to
calculate the calibration factor for determining N.
- Weigh samples and transfer to autosampler tray. Weigh a
second subsample to determine laboratory dry matter.
- Run samples.
- System must be capable of measuring nitrogen in feed
materials containing 0.2 to 20% nitrogen.
- Suitable fineness of grind is that which gives relative
standard deviation (RSD) £2.0% for 10 successive
determinations of nitrogen in mixture of corn grain and
soybeans (2/3and 1/3) that has been ground for analysis.
RSD, % = (standard deviation divided by mean %N) times
100. Fineness of grind (about 0.5 mm) required to achieve
this precision must be used for all mixed feeds and other
Calculation: Percent Nitrogen (N)
% N (DM basis) =% N (from analyzer output) Lab
Calculation: Percent Crude Protein (CP)
CP (DM basis)= % N (DM basis) X F
- F = 6.25 for all forages and feeds except wheat grains
- F = 5.70 for wheat grains
Include a reagent blank, one sample of NIST SRM Uric Acid 913,
and one or more quality control (QC) samples in each run,
choosing QC samples by matching analyte levels and matrices of QC
samples to the samples in the run. Include at least one set of
duplicates in each run if single determinations are being made.
Accuracy of system is demonstrated by making 10 successive
determinations of nitrogen in nicotinic acid and 10 successive
determinations in lysine.monohydrochloride. Means of
determinations must be within ±0.15 of the respective
theoretical values, with standard deviations £0.15. Standard
tryptophan may be substituted for lysine.monohydrochloride.
An acceptable average standard deviation among replicated
analyses for crude protein ranges from about ±0.10 for samples
with 10% CP to ±0.20 for samples with 20% CP, which results in
warning limits (2s) ranging from ±0.20 to 0.40 and control
limits (3s) ranging from ±0.30 to 0.60. Plot the results of the
control sample(s) on a X-control chart and examine the chart for
trends. Results outside of upper or lower warning limits, ±2s
(95 percent confidence limits), are evidence of possible problems
with the analytical system. Results outside of upper or lower
control limits, ±3s (99 percent confidence limits), indicate
loss of control and results of the run should be discarded. Two
consecutive analyses falling on one side of the mean between the
warning limits and the control limits also indicate loss of